# REELD — complete site text This file contains the full readable text of every page on reeld.org, in one document, for search engines and LLM crawlers. Canonical index: https://reeld.org/llms.txt Generated from 43 pages. ## Page index - Rare variant disease biology | REELD — https://reeld.org/ - Rare variant disease biology | REELD — https://reeld.org/ - Research | REELD — https://reeld.org/research - Studies | REELD — https://reeld.org/studies - Case Studies | REELD — https://reeld.org/case-studies - Editorials | REELD — https://reeld.org/editorials - Insights | REELD — https://reeld.org/insights - Methods | REELD — https://reeld.org/methods - Data practice | REELD — https://reeld.org/data - Library | REELD — https://reeld.org/library - Collaborate | REELD — https://reeld.org/collaborate - About | REELD — https://reeld.org/about - Our Story | REELD — https://reeld.org/team - News | REELD — https://reeld.org/news - REELD will exhibit at ASHG 2026 in Montréal | REELD — https://reeld.org/news/ashg-2026 - Contact | REELD — https://reeld.org/contact - Transparency | REELD — https://reeld.org/transparency - Privacy | REELD — https://reeld.org/privacy - Cell-state restriction clarifies MYH7 variant signals | REELD — https://reeld.org/studies/cell-state-restriction-myh7 - Convergent phenotype modules across neurodevelopmental genes | REELD — https://reeld.org/studies/convergent-neurodevelopmental-modules - Constraint-aware prioritization in pediatric cardiomyopathy | REELD — https://reeld.org/studies/constraint-aware-cardiomyopathy - Cross-atlas replication of dosage-sensitive programs | REELD — https://reeld.org/studies/cross-atlas-dosage-programs - Ontology depth changes gene-ranking stability | REELD — https://reeld.org/studies/ontology-depth-ranking-stability - Rare variants and microglial activation states | REELD — https://reeld.org/studies/microglial-activation-states - A reproducible evidence stack for inherited cardiomyopathy | REELD — https://reeld.org/case-studies/cardiomyopathy-evidence-stack - Finding a developmental window in epilepsy genetics | REELD — https://reeld.org/case-studies/developmental-epilepsy-cell-window - Negative controls sharpen an immune-dysregulation signal | REELD — https://reeld.org/case-studies/immune-dysregulation-negative-controls - Reconciling multisystem phenotypes in mitochondrial disease | REELD — https://reeld.org/case-studies/mitochondrial-phenotype-reconciliation - Reproducibility is a threshold, not a slogan | REELD — https://reeld.org/editorials/reproducibility-is-a-threshold - A phenotype ontology is part of the model | REELD — https://reeld.org/editorials/phenotype-ontology-is-a-model - Public data deserves primary-evidence discipline | REELD — https://reeld.org/editorials/public-data-is-primary-evidence - From target nomination to therapeutic hypothesis | REELD — https://reeld.org/editorials/from-target-to-hypothesis - Publish the negative space around a result | REELD — https://reeld.org/editorials/publish-the-negative-space - Choosing a cell atlas without choosing your conclusion | REELD — https://reeld.org/insights/choosing-a-cell-atlas - What gene constraint can and cannot say | REELD — https://reeld.org/insights/what-gene-constraint-can-say - When an ontology release changes your result | REELD — https://reeld.org/insights/ontology-release-drift - Building a negative-control signature library | REELD — https://reeld.org/insights/negative-control-library - Why we prefer an evidence graph to one score | REELD — https://reeld.org/insights/evidence-graph-not-score - A minimum provenance record for public data | REELD — https://reeld.org/insights/public-data-provenance - Variant to phenotype | REELD — https://reeld.org/research/variant-to-phenotype - Single-cell context | REELD — https://reeld.org/research/single-cell-context - Phenotype ontology | REELD — https://reeld.org/research/phenotype-ontology - Therapeutic hypotheses | REELD — https://reeld.org/research/therapeutic-hypotheses --- # Rare variant disease biology | REELD URL: https://reeld.org/ Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Rare disease computational biology Rare variants. Clearer disease biology. We connect genomic variation to cell states, phenotypes, and testable therapeutic hypotheses. Explore research Read studies One disease mechanism, tested from four directions. We build evidence that can survive a new dataset, a changed parameter, and a skeptical reader. Variant to phenotype Connect variant consequence to structured clinical observations without collapsing uncertainty. Single-cell context Locate disease mechanisms in the cell types and states where they become biologically coherent. Phenotype ontology Treat phenotype representation as a tested, versioned part of the analytical model. Therapeutic hypotheses Translate convergent evidence into falsifiable intervention logic and experimental priorities. Featured study Convergent phenotype modules across neurodevelopmental genes Testing whether distinct rare-disease genes converge on shared developmental programs. Question Do genes linked to overlapping neurodevelopmental phenotypes converge in the same developmental cell programs? Study type Cross-atlas methods demonstration Read the study From association to an experiment worth running. Every REELD hypothesis carries its evidence path, competing explanations, and conditions for failure. See our methods Variant Cell state Phenotype Hypothesis Reproducibility Independent references and sensitivity tests. Provenance Versioned inputs and auditable transformations. Falsifiability Predictions paired with ways they can fail. Case studies See how the framework behaves when the evidence is incomplete, heterogeneous, and sensitive to context. Cardiovascular disease A reproducible evidence stack for inherited cardiomyopathy From heterogeneous variant annotations to a testable cardiomyocyte-state hypothesis. View case study Neurodevelopment Finding a developmental window in epilepsy genetics Using cortical atlases to locate when a phenotype-linked gene module is most coherent. View case study All case studies Critical review is part of the research. Our editorials examine the assumptions that decide whether a computational result can travel. All editorials REELD Editorial August 19, 2026 Reproducibility is a threshold, not a slogan A mechanism that disappears under a reasonable alternate pipeline is not ready to anchor a therapeutic hypothesis. 18 min read Methods Editorial July 22, 2026 A phenotype ontology is part of the model Terms, ancestors, and information content shape the answer. They belong in the methods, not in a footnote. 16 min read Data Stewardship Editorial June 14, 2026 Public data deserves primary-evidence discipline Secondary analysis is not second-class science, but it must respect cohort design, provenance, and consent boundaries. 19 min read Built on our own research, engaged with leading campuses. Our analyses draw on our own research and findings, alongside engagement with investigators and scholarship at leading research campuses. How we collaborate Stanford University MIT The University of Texas Broad Institute Harvard Medical School UC San Diego REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Rare variant disease biology | REELD URL: https://reeld.org/ Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Rare disease computational biology Rare variants. Clearer disease biology. We connect genomic variation to cell states, phenotypes, and testable therapeutic hypotheses. Explore research Read studies One disease mechanism, tested from four directions. We build evidence that can survive a new dataset, a changed parameter, and a skeptical reader. Variant to phenotype Connect variant consequence to structured clinical observations without collapsing uncertainty. Single-cell context Locate disease mechanisms in the cell types and states where they become biologically coherent. Phenotype ontology Treat phenotype representation as a tested, versioned part of the analytical model. Therapeutic hypotheses Translate convergent evidence into falsifiable intervention logic and experimental priorities. Featured study Convergent phenotype modules across neurodevelopmental genes Testing whether distinct rare-disease genes converge on shared developmental programs. Question Do genes linked to overlapping neurodevelopmental phenotypes converge in the same developmental cell programs? Study type Cross-atlas methods demonstration Read the study From association to an experiment worth running. Every REELD hypothesis carries its evidence path, competing explanations, and conditions for failure. See our methods Variant Cell state Phenotype Hypothesis Reproducibility Independent references and sensitivity tests. Provenance Versioned inputs and auditable transformations. Falsifiability Predictions paired with ways they can fail. Case studies See how the framework behaves when the evidence is incomplete, heterogeneous, and sensitive to context. Cardiovascular disease A reproducible evidence stack for inherited cardiomyopathy From heterogeneous variant annotations to a testable cardiomyocyte-state hypothesis. View case study Neurodevelopment Finding a developmental window in epilepsy genetics Using cortical atlases to locate when a phenotype-linked gene module is most coherent. View case study All case studies Critical review is part of the research. Our editorials examine the assumptions that decide whether a computational result can travel. All editorials REELD Editorial August 19, 2026 Reproducibility is a threshold, not a slogan A mechanism that disappears under a reasonable alternate pipeline is not ready to anchor a therapeutic hypothesis. 18 min read Methods Editorial July 22, 2026 A phenotype ontology is part of the model Terms, ancestors, and information content shape the answer. They belong in the methods, not in a footnote. 16 min read Data Stewardship Editorial June 14, 2026 Public data deserves primary-evidence discipline Secondary analysis is not second-class science, but it must respect cohort design, provenance, and consent boundaries. 19 min read Built on our own research, engaged with leading campuses. Our analyses draw on our own research and findings, alongside engagement with investigators and scholarship at leading research campuses. How we collaborate Stanford University MIT The University of Texas Broad Institute Harvard Medical School UC San Diego REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Research | REELD URL: https://reeld.org/research Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Research Mechanisms that remain visible under pressure. REELD integrates rare-variant evidence, single-cell context, and structured phenotypes to produce hypotheses that can be challenged. Variant to phenotype Connect variant consequence to structured clinical observations without collapsing uncertainty. Variant consequence, gene constraint, phenotype fit Explore program Single-cell context Locate disease mechanisms in the cell types and states where they become biologically coherent. Cell identity, state, development, perturbation Explore program Phenotype ontology Treat phenotype representation as a tested, versioned part of the analytical model. Term specificity, propagation, semantic similarity Explore program Therapeutic hypotheses Translate convergent evidence into falsifiable intervention logic and experimental priorities. Direction of effect, intervention point, validation logic Explore program Our research standard Confidence is earned through convergence, not visual complexity. Define the claim State the mechanism and its plausible alternatives. Build the evidence graph Keep each source and inference independently visible. Stress test the result Change references, parameters, and negative controls. Specify the experiment Translate stable evidence into a falsifiable next step. REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Studies | REELD URL: https://reeld.org/studies Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Studies Open analyses of rare-disease mechanisms. Methods-led research built from public data, versioned assumptions, and explicit limitations. All Cardiovascular genetics Neurodevelopment Functional genomics Phenotype informatics Neuroimmunology Methods study Cardiovascular genetics Cell-state restriction clarifies MYH7 variant signals A reproducible workflow for asking when cardiomyopathy-associated variation becomes biologically legible. Read study Atlas study Neurodevelopment Convergent phenotype modules across neurodevelopmental genes Testing whether distinct rare-disease genes converge on shared developmental programs. Read study Benchmark study Cardiovascular genetics Constraint-aware prioritization in pediatric cardiomyopathy Integrating population constraint without allowing it to overwhelm tissue and phenotype evidence. Read study Replication study Functional genomics Cross-atlas replication of dosage-sensitive programs A study of which haploinsufficiency signals persist across public single-cell references. Read study Ontology study Phenotype informatics Ontology depth changes gene-ranking stability Measuring how broad and specific phenotype terms alter computational prioritization. Read study Mechanism study Neuroimmunology Rare variants and microglial activation states Separating disease-associated state signatures from generalized inflammatory response. Read study REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Case Studies | REELD URL: https://reeld.org/case-studies Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Case studies What the framework does with difficult evidence. Worked research scenarios show the decisions, controls, and caveats behind a credible mechanism hypothesis. Cardiovascular disease A reproducible evidence stack for inherited cardiomyopathy From heterogeneous variant annotations to a testable cardiomyocyte-state hypothesis. View case study Neurodevelopment Finding a developmental window in epilepsy genetics Using cortical atlases to locate when a phenotype-linked gene module is most coherent. View case study Immunogenomics Negative controls sharpen an immune-dysregulation signal Testing a rare-disease module against common inflammatory and technical signatures. View case study Metabolic genetics Reconciling multisystem phenotypes in mitochondrial disease Using ontology structure to preserve organ-specific evidence without fragmenting the case. View case study Scope note These are editorialized methods demonstrations based on public-data research patterns. They are not patient reports, clinical validations, or claims of institutional partnership. REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Editorials | REELD URL: https://reeld.org/editorials Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Editorials Peer-review thinking, published in the open. Critical essays on evidence quality, reproducibility, data stewardship, and translational inference. REELD editorials are institutional perspectives, not peer-reviewed journal articles. They apply the standards we expect during scientific review. REELD Editorial August 19, 2026 Reproducibility is a threshold, not a slogan A mechanism that disappears under a reasonable alternate pipeline is not ready to anchor a therapeutic hypothesis. 18 min read Methods Editorial July 22, 2026 A phenotype ontology is part of the model Terms, ancestors, and information content shape the answer. They belong in the methods, not in a footnote. 16 min read Data Stewardship Editorial June 14, 2026 Public data deserves primary-evidence discipline Secondary analysis is not second-class science, but it must respect cohort design, provenance, and consent boundaries. 19 min read Translational Editorial May 28, 2026 From target nomination to therapeutic hypothesis A ranked gene is the start of the argument. Cell context, direction of effect, and intervention logic complete it. 18 min read Open Science Editorial April 9, 2026 Publish the negative space around a result Failed replications, null tissues, and unstable rankings define where a mechanism does not travel. 15 min read Our review lens Evidence Does the claim follow from the data, and are alternate explanations tested? Reproducibility Can another group reconstruct inputs, transformations, and analytical choices? Translation Is the direction of effect, cellular context, and next experiment explicit? REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Insights | REELD URL: https://reeld.org/insights Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Insights Notes from the analytical work. Practical guidance on public data, model assumptions, and the details that decide whether a result reproduces. Search insights Methods August 11, 2026 Choosing a cell atlas without choosing your conclusion A practical checklist for tissue coverage, donor structure, annotation depth, and replication. Read note Rare variants July 31, 2026 What gene constraint can and cannot say Constraint is strong population evidence, but it is not a disease mechanism by itself. Read note Phenotypes July 3, 2026 When an ontology release changes your result Version drift can alter term relationships and ranking behavior. Here is how we audit it. Read note Single cell June 18, 2026 Building a negative-control signature library Common stress, cell-cycle, interferon, and dissociation programs belong in every enrichment workflow. Read note Interpretation May 16, 2026 Why we prefer an evidence graph to one score One number hides disagreement. A graph preserves which evidence supports each step. Read note Open science April 24, 2026 A minimum provenance record for public data Source accessions, versions, transformations, exclusions, and checks every analysis should retain. Read note REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Methods | REELD URL: https://reeld.org/methods Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Methods An analysis is only as strong as its audit trail. REELD workflows are designed to be reconstructed, challenged, and transferred across public datasets. Ingest with provenance Every source enters with accession, release, licensing context, cohort metadata, and file identity. Harmonize explicitly Gene identifiers, cell labels, phenotype terms, and reference builds are mapped with retained ambiguity. Model the question Analyses are built around a named mechanism and its plausible competing explanations. Integrate evidence Genomic, cellular, and phenotype layers remain visible inside one evidence graph. Stress test We vary datasets, parameters, ontology rules, controls, and label resolution. Release reproducibly Code, environment locks, manifests, and limitations travel with every release-ready output. What accompanies a release-ready analysis A result without its analytical context is not portable. Source manifest Accessions, versions, licenses, and exclusions Environment lock Packages, references, and execution context Sensitivity report Alternate parameters, datasets, and negative controls Interpretation memo Claim, uncertainty, alternatives, and next experiment Need a methods review? We can help pressure-test a rare-disease analysis before it becomes a mechanistic claim. Propose a collaboration REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Data practice | REELD URL: https://reeld.org/data Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Data practice Public data, with its history intact. We preserve where data came from, what changed, and where reuse should stop. The source is part of the result. Cohort design, reference version, tissue handling, and access terms remain attached to every derived object. Genomic references Population variation, gene constraint, transcript consequences, and disease association resources. Single-cell atlases Public tissue, organoid, developmental, and disease references with donor-aware analysis. Phenotype knowledge Human Phenotype Ontology and disease-gene resources with explicit release tracking. Functional context Pathway, interaction, perturbation, and expression resources used as supporting evidence. Data stewardship Minimum necessary use We use only the fields and resolution required for the research question. Source-aligned terms Licenses, controlled-access conditions, and attribution requirements travel downstream. No re-identification REELD does not attempt to identify participants in public or controlled datasets. Careful communication Results are framed to avoid stigmatizing populations or overstating clinical meaning. Useful public resources Examples of the infrastructure our field relies on. Resource use depends on each source's current terms. gnomAD Human Cell Atlas Human Phenotype Ontology NCBI GEO REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Library | REELD URL: https://reeld.org/library Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Research library Studies, cases, methods, and critical perspective. The complete REELD collection, organized by the role each item plays in the evidence process. Studies 6 research outputs Methods study Cell-state restriction clarifies MYH7 variant signals Atlas study Convergent phenotype modules across neurodevelopmental genes Benchmark study Constraint-aware prioritization in pediatric cardiomyopathy Replication study Cross-atlas replication of dosage-sensitive programs Ontology study Ontology depth changes gene-ranking stability Mechanism study Rare variants and microglial activation states Case studies 4 methods demonstrations Cardiovascular disease A reproducible evidence stack for inherited cardiomyopathy Neurodevelopment Finding a developmental window in epilepsy genetics Immunogenomics Negative controls sharpen an immune-dysregulation signal Metabolic genetics Reconciling multisystem phenotypes in mitochondrial disease Editorials 5 perspectives August 19, 2026 Reproducibility is a threshold, not a slogan July 22, 2026 A phenotype ontology is part of the model June 14, 2026 Public data deserves primary-evidence discipline May 28, 2026 From target nomination to therapeutic hypothesis April 9, 2026 Publish the negative space around a result Insights 6 practical notes Methods Choosing a cell atlas without choosing your conclusion Rare variants What gene constraint can and cannot say Phenotypes When an ontology release changes your result Single cell Building a negative-control signature library Interpretation Why we prefer an evidence graph to one score Open science A minimum provenance record for public data REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Collaborate | REELD URL: https://reeld.org/collaborate Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Collaborate Bring a question that needs more than one dataset. We work with research groups that value transparent inference, robust negative controls, and a clear experimental handoff. Strong fit Projects with a focused disease question, usable public evidence, and a path to experimental or clinical-domain review. Mechanism mapping Place a candidate gene or program in its relevant cell and phenotype context. Reanalysis Test whether a published or internal result transfers across public references. Methods review Audit ontology, enrichment, or evidence-integration choices before publication. Hypothesis design Turn a stable computational signal into falsifiable experimental logic. Across the academic research ecosystem REELD's research draws on our own findings alongside engagement with investigators and laboratories at leading research campuses, including Stanford, MIT, and The University of Texas. Institutional clarity Naming an institution reflects engagement with its research or investigators. It does not imply a formal, exclusive, or ongoing partnership with that institution. Start a conversation How an engagement begins Research question A concise claim, disease context, and decision the analysis should inform. Evidence audit A review of available data, access conditions, confounders, and domain expertise. Analysis charter Predefined outputs, robustness tests, boundaries, authorship, and release expectations. REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # About | REELD URL: https://reeld.org/about Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate About REELD A computational lab for rare-disease mechanism discovery. We turn fragmented public evidence into reproducible biological arguments and practical experimental priorities. Our premise Rare disease is a test of how carefully science can connect evidence. A variant may be individually rare while the affected cellular program is shared across diseases. Public genomic, single-cell, and phenotype resources now make those shared mechanisms more visible. REELD exists to find that convergence without erasing uncertainty. We combine computational depth with a strict standard for provenance, replication, and interpretive boundaries. REELD is a non-profit educational research institute. Evidence before narrative We build the argument from independently inspectable layers. Negative results have structure A failed transfer defines the boundary of a mechanism. Public work should travel Releases are designed for reuse, criticism, and extension. Translation needs specificity Cell context and direction of effect matter as much as target identity. “A credible hypothesis should become clearer when its assumptions are exposed.” REELD research principle Read the work. Challenge the assumptions. Explore the library Review our standards REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Our Story | REELD URL: https://reeld.org/team Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Our Story A non-profit institute, built to advance the world's understanding of biology. REELD exists for one reason: to further human understanding of biology — genetics, cellular systems, and the mechanisms of disease — for the good of everyone it touches. Our premise Why we exist REELD is a non-profit educational research institute, founded in 2016. We are not built to generate proprietary advantage or shareholder return. We are built to generate understanding — of rare genetic variation, single-cell biology, phenotype ontology, and the disease mechanisms that connect them — and to put that understanding into the world where it can be used, tested, and built upon. At the center of our mission is a simple belief: biology is too important, and too difficult, for any one lab to understand alone. Progress comes from shared evidence, shared scrutiny, and shared standards. Rare-variant genetics How individually rare genetic variants converge on shared disease mechanisms. Single-cell biology The cell types, cell states, and developmental windows where disease mechanisms become visible. Phenotype ontology How clinical observations can be represented, compared, and reasoned about rigorously. Translational biology Turning convergent evidence into falsifiable hypotheses that can guide real experiments. Research alongside the academic world REELD's research draws on our own findings, alongside engagement with the work of investigators and laboratories at leading research universities, including Stanford, MIT, and The University of Texas. Naming an institution reflects engagement with its research or scholars — not a formal or exclusive partnership. Bigger than any one institution This engagement is deliberate. The questions we study are bigger than any single institution, and the understanding we're after belongs to the field, not to us alone. Our principal investigators are computational biologists, human geneticists, and clinician-scientists. Start a conversation Why non-profit We chose this structure so we could work on what actually matters. Non-profit was a deliberate choice, not a technicality. A commercial lab answers to a return. That pressure decides which questions get asked: the ones with a near-term market, a defensible asset, a path to revenue. Enormous amounts of important biology fall outside those lines — rare conditions affecting too few people to justify a business case, negative results that save the field years, methods work that makes everyone's science better without belonging to anyone. That work is exactly what we exist to do. Being a non-profit means we can spend a year on a mechanism that may turn out to be a dead end, and publish it anyway, because the field is better for knowing. It means we can study a disease because it is poorly understood, not because it is profitable. It means our results can be given away rather than licensed. We chose this structure because it is the only one that lets us follow the science where it actually leads. Our mission At the forefront of the craft, for the good of mankind Our mission has two parts, and neither stands without the other. First, to advance the craft itself — the methods, the standards of evidence, the discipline of reproducibility — so that biological research becomes more rigorous, more transparent, and more trustworthy over time. Second, to advance the world's understanding of biology in service of human health — because every mechanism we clarify, every hypothesis we sharpen, is a step toward therapies and knowledge that can help people. We do this work for the good of mankind. That is not a tagline — it is the standard we hold every study, case study, and editorial to. “Understanding belongs to the field, not to any one institute.” REELD research principle Read the work. Join the mission. Explore our research Collaborate with us REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # News | REELD URL: https://reeld.org/news Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate News Announcements from the institute. Announcement · Meetings August 28, 2026 REELD will exhibit at ASHG 2026 in Montréal REELD will host a booth for the full duration of the 76th Annual Meeting of the American Society of Human Genetics, October 20–24, 2026, at the Palais des Congrès de Montréal. Read announcement REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # REELD will exhibit at ASHG 2026 in Montréal | REELD URL: https://reeld.org/news/ashg-2026 Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Announcement · Meetings REELD will exhibit at ASHG 2026 in Montréal REELD will host a booth for the full duration of the 76th Annual Meeting of the American Society of Human Genetics, October 20–24, 2026, at the Palais des Congrès de Montréal. Where to find us REELD will be present in the exhibit hall for all five days of the meeting, October 20 through October 24. Members of our research team will be at the booth throughout, and we welcome researchers, clinicians, students, and prospective collaborators to stop by at any point during exhibit hours. Why we are attending ASHG is the principal annual meeting of the human genetics community, and much of the methodological discussion relevant to our work takes place there: rare-variant interpretation, single-cell disease context, phenotype-ontology analysis, and the evidentiary standards applied to translational claims. We are attending to share how we approach these problems, to hear how other groups are addressing them, and to identify opportunities for collaboration. We welcome critical examination of our methods and are particularly interested in speaking with investigators working on related questions. What we would like to talk about We would be glad to discuss: Rare-variant interpretation and the limits of constraint-based prioritization Single-cell context and cell-state restriction of disease mechanisms Phenotype-ontology methodology, propagation, and ranking stability Reproducibility standards and cross-dataset replication Negative results, and what they define about the boundary of a mechanism Potential collaborations with investigators and laboratories If you are working on a mechanism question that has become difficult to resolve, bring it. Those are our favorite conversations. Meet with us To arrange a specific time to meet during the meeting, or to let us know you are coming, write to lab@reeld.org with “ASHG 2026” in the subject line. We will do our best to accommodate scheduled conversations around exhibit hours. We look forward to meeting our fellow researchers in Montréal. Meeting 76th Annual Meeting of the American Society of Human Genetics (ASHG 2026) Dates Tuesday, October 20 – Saturday, October 24, 2026 Venue Palais des Congrès de Montréal 201 Av. Viger O Montréal, Quebec H2Z 1X7 Canada Booth To be announced Arrange a meeting REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Contact | REELD URL: https://reeld.org/contact Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Contact Start with the research question. Tell us what you are trying to explain, which evidence exists, and what decision the analysis should support. Research inquiries lab@reeld.org Include only non-sensitive information in your first message. Do not send patient-level data or protected health information. Good first note Disease or phenotype context Available public datasets Primary analytical uncertainty Desired experimental handoff Name Email Institution or organization Optional Research question At least 20 characters. Please do not include sensitive data. Prepare email REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Transparency | REELD URL: https://reeld.org/transparency Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Transparency What we claim, what we do not, and what should be verifiable. Scientific credibility depends on boundaries that are as visible as the positive result. Research status Studies on this site are labeled by output type. Methods demonstrations and research briefs should not be read as peer-reviewed clinical evidence. Institution references University names identify institutions and investigators whose research or scholarship informs our work. Naming an institution does not imply a formal partnership or endorsement. Public data REELD follows source access conditions, retains provenance, and avoids participant re-identification. Data terms remain authoritative. Clinical boundary REELD content supports research prioritization. It is not medical advice, diagnostic interpretation, or a substitute for accredited clinical genetics practice. Certifications REELD maintains multiple organizational and professional certifications. Credential names, issuing bodies, scopes, and current status are supplied during institutional diligence for direct verification. Standard limitations Public cohorts can underrepresent populations, tissues, ages, and disease stages. Single-cell references vary in sample handling, depth, annotation, and donor structure. Phenotype records can reflect ascertainment, documentation, and ontology-version bias. Computational convergence prioritizes a mechanism for testing. It does not establish causality. REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Privacy | REELD URL: https://reeld.org/privacy Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Privacy A minimal-data website. This production-ready policy is a starting point and should be reviewed against the analytics, hosting, and email services selected for deployment. Information you provide If you contact REELD by email, we receive the information you include. Do not send patient-level data, protected health information, credentials, or unpublished sensitive datasets through the website. Newsletter form The current site demonstrates subscription interaction locally and does not transmit the entered address. Connect the form to an approved provider and update this notice before collecting subscriptions. Technical data The site does not include third-party analytics by default. A hosting provider may process standard request logs for security and reliability under its own terms. Contact Questions about privacy can be sent to lab@reeld.org . REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Cell-state restriction clarifies MYH7 variant signals | REELD URL: https://reeld.org/studies/cell-state-restriction-myh7 Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All studies Methods study August 2026 Open analysis Cell-state restriction clarifies MYH7 variant signals A reproducible workflow for asking when cardiomyopathy-associated variation becomes biologically legible. REELD research visualization. Image is illustrative and does not represent a measured result. Research question Can cell-type and cell-state context separate broadly expressed gene signal from disease-relevant cardiac biology? Evidence layers Transcript-aware variant annotation Donor-aware cardiac single-cell analysis Developmental state mapping Phenotype-ontology propagation Matched null gene sets Leave-one-atlas-out replication Technical profile Biological unit Cardiomyocyte state and sarcomere program Primary contrast State-restricted signal versus broad cardiac expression Inference level Mechanism prioritization, not variant classification Decisive control Independent-atlas and matched-gene-set replication Primary analytical observation MYH7 becomes most interpretable when variant consequence, sarcomere biology, ventricular cardiomyocyte state, and cardiac phenotype agree. Broad tissue expression alone is insufficient, and signals that depend on a single atlas or label resolution should not be promoted to mechanism claims. Executive interpretation At a high level, this study asks a simple question: when a rare MYH7 variant is biologically important, where should its effect become visible? A conventional tissue-level answer—“the heart”—is correct but not discriminating. The heart contains multiple cell classes, and cardiomyocytes themselves occupy developmental, metabolic, contractile, and stress-associated states that can make the same gene signal mean different things. Our conclusion is that cellular context should be treated as an evidence gate. A candidate mechanism is strengthened when the implicated variant consequence is compatible with established MYH7 biology, the relevant phenotype is cardiac and structurally coherent, and the signal localizes to reproducible ventricular cardiomyocyte programs. It is weakened when enrichment is driven by ubiquitous expression, one annotation source, one donor, or one atlas taxonomy. Biological and genetic context MYH7 encodes beta-myosin heavy chain, a major component of the thick filament in cardiac sarcomeres. Pathogenic variation has established relationships with inherited cardiomyopathies, but the molecular direction and phenotypic expression are not interchangeable across variant classes. Missense variation affecting the motor or filament-forming regions cannot be interpreted as though it were generic loss of function, and a gene-level association cannot substitute for transcript, domain, inheritance, segregation, or variant-specific evidence. This distinction matters computationally. Expression-based analyses often reward genes that are abundant, well annotated, and connected to many known pathways. MYH7 satisfies all three conditions. Without explicit null models, a strong score can therefore recapitulate prior knowledge rather than identify the cell state in which a proposed mechanism is most testable. We frame the analysis around localization and coherence, not rediscovery of cardiac expression. Separate gene–disease validity from variant-level pathogenicity. Retain transcript and protein-domain context rather than collapsing to one gene symbol. Model hypertrophic, dilated, skeletal-muscle, and mixed phenotypic branches independently before testing convergence. Evidence assembly and harmonization The input layer combines public variant annotations, population-frequency evidence, gene- and transcript-level constraint, structured phenotypes, and cardiac single-cell or single-nucleus references. Every derived object retains its source release, genome build, transcript set, identifier mapping, exclusion rules, and download date. Conflicting consequences are preserved as disagreement rather than resolved silently. Cardiac atlases are harmonized twice. First, original author labels are retained so the analysis respects the resolution supported by each study. Second, labels are mapped to a conservative shared hierarchy: non-myocyte compartment, cardiomyocyte lineage, chamber-enriched identity, maturation state, and stress or remodeling state. Results must be intelligible at both the source-label and shared-label levels. A conclusion that exists only after aggressive relabeling is considered taxonomy-dependent. Cell-state model and statistical design Cells are not treated as independent biological replicates. Where donor identifiers and counts are available, expression is aggregated within donor-by-state strata to create pseudobulk profiles, and donor is the unit of inference. This reduces pseudoreplication and prevents a large cell yield from one specimen from masquerading as biological certainty. Detection fraction, normalized abundance, rank-based specificity, and module coherence are reported separately because each answers a different question. The primary state score is an evidence vector rather than a single opaque number. It contains transcript-compatible expression, state specificity, co-expression with sarcomere and proteostasis modules, phenotype concordance, and cross-reference transfer. Alternate weights can reorder candidates without changing the underlying evidence graph. This makes it possible to ask whether the conclusion survives a reasonable change in priorities rather than defending one preferred parameterization. Donor-stratified pseudobulk summaries for inferential comparisons. Matched null genes selected on expression level, transcript length, and constraint neighborhood. Bootstrap intervals over donors and leave-one-reference-out replication over atlases. Sensitivity to chamber labels, maturation resolution, and ontology propagation depth. Controls and competing explanations The main competing explanation is abundance: MYH7 may score highly because it is a dominant cardiac transcript, not because the tested cell state is mechanistically informative. We address this with expression-matched null genes and by requiring enrichment beyond a broad cardiomyocyte baseline. A second explanation is annotation circularity, in which known cardiomyopathy labels enter both the phenotype definition and the validation set. We therefore maintain phenotype-only, cell-state-only, and integrated analyses and inspect their agreement. A third explanation is atlas-specific composition. Adult surgical tissue, developmental tissue, organoid references, and diseased myocardium sample different biology. We do not expect identical effect sizes across these resources. We require agreement in direction and biological neighborhood, while allowing the most specific label to differ. Failure to transfer is reported as a boundary of the hypothesis rather than erased by integration. Technical findings Across the analytical variants we consider credible, the informative signal is not “MYH7 is expressed in heart.” The stable observation is that MYH7-linked evidence remains coherent in ventricular cardiomyocyte programs characterized by organized contractile machinery, sarcomere maintenance, force generation, and maturation-associated metabolic support. Broad fibroblast, endothelial, immune, and undifferentiated progenitor compartments do not carry equivalent integrated support after expression matching. The result is sensitive to mechanism specification. A missense-compatible contractile hypothesis remains interpretable when evaluated in mature sarcomere-bearing states, whereas a generic haploinsufficiency model is not supported simply by high cardiac expression. Phenotype terms that distinguish ventricular hypertrophy, chamber dilation, impaired systolic function, or skeletal-muscle involvement materially change which mechanistic branch is coherent. This is a desired behavior: the model should respond to biologically informative phenotype differences. Consolidated findings What the analysis establishes Cell state adds discrimination Ventricular cardiomyocyte maturation and contractile programs carry more mechanistic information than organ-level expression or cell-type detection alone. Mechanism cannot be inferred from abundance High MYH7 expression establishes relevance to contractile tissue but does not determine gain of function, dominant-negative behavior, reduced motor performance, or pathogenicity. Replication is biological, not lexical The most stable evidence transfers as a conserved sarcomere-centered neighborhood even when atlas-specific cell-state names differ. Phenotype structure changes the branch Hypertrophic, dilated, and skeletal-muscle features should be represented as partially shared but non-equivalent phenotype programs. Research conclusion Conclusion and experimental handoff We conclude that an MYH7 variant hypothesis is ready for experimental prioritization only when it specifies the variant mechanism, the cardiomyocyte state, the phenotype branch, and a directional molecular readout. The most defensible context from this analysis is a mature or maturing ventricular cardiomyocyte system in which sarcomere organization, contractile kinetics, force production, and energetic adaptation can be measured together. The direct next step is not another generic enrichment analysis. It is a variant-specific perturbation in an isogenic cardiomyocyte model or engineered cardiac tissue, with rescue or correction, matched differentiation state, and orthogonal readouts of myosin function and cellular mechanics. A result that fails to localize to the predicted state—or that persists identically in non-cardiac controls—would weaken the proposed context-specific mechanism. Interpretive boundary Limitations Public atlases are observational and differ in donor composition, tissue acquisition, platform, and annotation depth. Single-cell RNA abundance does not directly measure myosin protein stoichiometry, motor kinetics, sarcomere ultrastructure, or tissue-level force. The framework prioritizes experimental context; it does not classify a variant or estimate penetrance. Developmental and disease-remodeling states are incompletely sampled and may not transfer cleanly to in-vitro cardiomyocytes. Technical vocabulary Glossary Pseudobulk Aggregation of counts within a biological replicate and cell state so the replicate—not each cell—is the inferential unit. State specificity The degree to which a signal is concentrated in a defined cellular program rather than broadly detected. Matched null set Control genes selected to resemble target genes on nuisance properties such as expression and length. Evidence vector A set of independently inspectable evidence dimensions retained instead of compressed into one score. Reproducibility and evidentiary scope This study is a REELD public-data analysis and methods interpretation. It does not report a newly recruited clinical cohort, classify an individual variant, or replace clinical review. A release-ready execution of the workflow includes accession-level provenance, source and ontology versions, code state, environment locks, predefined sensitivity analyses, and output checksums. Continue exploring Browse all studies REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Convergent phenotype modules across neurodevelopmental genes | REELD URL: https://reeld.org/studies/convergent-neurodevelopmental-modules Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All studies Atlas study July 2026 Research brief Convergent phenotype modules across neurodevelopmental genes Testing whether distinct rare-disease genes converge on shared developmental programs. REELD research visualization. Image is illustrative and does not represent a measured result. Research question Do genes linked to overlapping neurodevelopmental phenotypes converge in the same developmental cell programs? Evidence layers Human Phenotype Ontology Developmental cortical atlases Information-content weighting Expression-matched resampling Donor-aware module scoring Cross-atlas transfer Technical profile Biological unit Developmental cortical cell program Primary contrast Phenotype-linked module versus matched gene sets Inference level Shared developmental context Decisive control Ontology-depth and gene-composition sensitivity Primary analytical observation Phenotype-defined gene modules become biologically interpretable only after broad clinical terms, gene-set composition, developmental timing, and replicate structure are controlled. Stable convergence is concentrated in developmental programs, not in one universal “neurodevelopmental” cell type. Executive interpretation Many rare neurodevelopmental disorders share terms such as global developmental delay, intellectual disability, hypotonia, seizures, or abnormal behavior. Those terms are clinically important, but they are also common across genetically distinct conditions. If genes are grouped by broad vocabulary alone, almost any neuronal atlas can produce an apparently plausible enrichment. We find that convergence is most credible when it appears at the level of a developmental program and survives both phenotype and gene-set perturbation. The analysis does not support a single cellular mechanism for all neurodevelopmental disease genes. It supports narrower modules that align with specific transitions such as progenitor proliferation, neuronal specification, migration, synaptogenesis, or excitatory-neuron maturation. Phenotype module construction Phenotype sets are represented as graphs, not bags of labels. Exact Human Phenotype Ontology terms are retained with onset, frequency, negation, and uncertainty when available. Ancestors are propagated for recall, while information-content weighting prevents very broad ancestors from dominating similarity. We compare Resnik-style most-informative-common-ancestor similarity with normalized alternatives and report where rankings depend on the metric. Gene modules are created from phenotype-to-disease and disease-to-gene relationships under explicit evidence filters. A gene can participate in more than one module, which is biologically realistic but statistically consequential. Module overlap is therefore quantified, and enrichment is compared with degree-matched and expression-matched null sets so heavily annotated genes do not win merely because the knowledge graph contains more edges for them. Developmental atlas analysis The cortical references are organized along two simultaneous axes: cell lineage and developmental progression. Broad labels such as radial glia, intermediate progenitor, excitatory neuron, inhibitory neuron, glia, and vascular cell support cross-atlas transfer. Finer labels capture regional identity, maturation stage, and transient states. We retain both scales because broad agreement with fine-state disagreement is itself informative. Module activity is evaluated with rank-based gene-set scoring and donor-aware pseudobulk summaries. Genes with extremely high abundance or broad housekeeping roles are capped or tested separately. We also calculate leading-edge membership: the subset of genes repeatedly responsible for a module score. A module that transfers only because one ubiquitous gene dominates is not treated as convergent biology. Robustness and null architecture Four perturbations are central to the design. We vary ontology propagation depth, remove one phenotype term at a time, resample the gene set while matching expression and annotation degree, and leave out each atlas or donor group in turn. Stability is evaluated by direction, cell-program neighborhood, leading-edge overlap, and rank concordance rather than by one significance threshold. We also test generic neuronal, ribosomal, cell-cycle, stress, and synaptic gene programs as competing explanations. A disease module must show information beyond these controls. This is particularly important for synaptic genes, which are numerous, highly studied, and broadly detected across maturing neurons. Technical findings Broad developmental-delay modules show high recall but low localization: they are distributed across multiple neuronal lineages and are sensitive to highly annotated genes. When the phenotype representation is refined with seizure type, onset, morphologic features, movement abnormalities, or specific cognitive and behavioral features, the modules become smaller and more developmentally resolved. The most reproducible convergence is observed as program-level agreement across stages of neuronal differentiation and maturation. Some modules localize to proliferative or early specification programs, while others become coherent after neuronal identity is established and synaptic machinery is assembled. The exact atlas label is less stable than the developmental ordering, which argues for interpreting convergence along trajectories rather than treating clusters as fixed biological entities. What would falsify the interpretation The shared-program hypothesis would be weakened if the module score vanished after removal of one broad phenotype term, if matched random sets produced equivalent localization, if the leading edge were dominated by one or two ubiquitous genes, or if the developmental ordering reversed across independent references. It would also be weakened if perturbation of representative genes produced unrelated cellular phenotypes under a matched experimental context. These failure conditions are part of the output. They distinguish a reusable mechanistic module from a retrospective narrative assembled around a visually compelling embedding. Consolidated findings What the analysis establishes Broad terms recover candidates but blur mechanism High-level phenotype ancestors are useful for recall but are too nonspecific to localize a developmental program on their own. Trajectory position transfers better than cluster name Independent atlases disagree in taxonomy more often than they disagree in the ordering from progenitor to differentiated neuronal programs. Leading-edge stability is essential A module is credible when a coherent subset of genes repeatedly carries the signal, not when a different dominant gene appears in every reference. Convergence is modular, not universal Distinct phenotype-defined gene groups align with different developmental windows; they should not be collapsed into one pan-neurodevelopmental mechanism. Research conclusion Conclusion and experimental handoff We conclude that phenotype similarity can identify shared neurodevelopmental biology, but only after ontology depth, gene-set composition, and developmental time are made explicit. The useful output is a set of bounded developmental modules with named leading-edge genes and failure conditions—not a claim that all genes producing similar clinical language share one pathway. The experimental handoff is a matched perturbation panel performed at the predicted developmental stage. Readouts should include cell-state progression, morphology, electrophysiology or network activity when appropriate, and module-specific molecular markers. Testing the same perturbation too early, too late, or in an unrelated lineage would not adequately evaluate the hypothesis. Interpretive boundary Limitations Developmental atlases incompletely sample spatial, temporal, and ancestry diversity. Disease-gene annotations are uneven and favor well-studied disorders and genes. Transcriptomic convergence does not establish identical molecular direction across genes. In-vitro developmental systems may not reproduce human cortical timing or cell-cell interactions. Technical vocabulary Glossary Information content A measure of term specificity based on how rarely an ontology term or its descendants occur in annotations. Leading edge The subset of genes that repeatedly contributes most strongly to a gene-set enrichment or activity score. Ontology propagation Expansion of an observed term to include defined ancestors in the ontology graph. Trajectory An inferred ordering of cellular states that approximates a biological progression rather than direct elapsed time. Reproducibility and evidentiary scope This study is a REELD public-data analysis and methods interpretation. It does not report a newly recruited clinical cohort, classify an individual variant, or replace clinical review. A release-ready execution of the workflow includes accession-level provenance, source and ontology versions, code state, environment locks, predefined sensitivity analyses, and output checksums. Continue exploring Browse all studies REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Constraint-aware prioritization in pediatric cardiomyopathy | REELD URL: https://reeld.org/studies/constraint-aware-cardiomyopathy Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All studies Benchmark study June 2026 Protocol complete Constraint-aware prioritization in pediatric cardiomyopathy Integrating population constraint without allowing it to overwhelm tissue and phenotype evidence. REELD research visualization. Image is illustrative and does not represent a measured result. Research question How should gene constraint be balanced against cardiac context in rare-variant prioritization? Evidence layers LOEUF and missense constraint Transcript-aware consequence Cardiac cell-state specificity HPO semantic similarity Weight-grid benchmarking Rank stability analysis Technical profile Biological unit Gene–variant–phenotype evidence chain Primary contrast Constraint-dominant versus balanced models Inference level Research ranking Decisive control Weight perturbation and prior ablation Primary analytical observation Constraint improves prioritization when treated as a calibrated population prior with uncertainty. It degrades interpretability when used as a hard gate or as a substitute for inheritance, phenotype fit, transcript relevance, and cardiac context. Executive interpretation Population constraint measures whether classes of variation are depleted relative to expectation. It is powerful evidence about selection, but it is not a diagnosis and it does not identify the affected tissue. In pediatric cardiomyopathy, a constraint-heavy ranking can appear sophisticated while largely reproducing a list of genes that are broadly dosage sensitive. Our benchmark shows that constraint is most useful as a prior whose influence is conditional on the proposed mechanism. Loss-of-function intolerance is relevant to haploinsufficiency hypotheses; regional missense constraint may be relevant to altered-protein mechanisms; neither should automatically override a poor phenotype match, an irrelevant transcript, incompatible inheritance, or absent cardiac context. Constraint metrics and their uncertainty We retain observed and expected counts, the upper confidence bound of the observed-to-expected ratio for predicted loss-of-function variation, and missense depletion metrics rather than importing only a percentile. Genes with few expected variants have less informative estimates, and transcript-level differences can be substantial. A binary “constrained/not constrained” threshold discards this uncertainty. Predicted loss-of-function annotations also require quality control. Terminal truncations, low-expression exons, rescue transcripts, mapping artifacts, and variants unlikely to trigger nonsense-mediated decay can inflate an apparent loss-of-function count. Constraint is therefore linked to transcript-aware consequence review and is down-weighted when the disease-relevant transcript or exon context is uncertain. Benchmark architecture Candidate genes are ranked under a grid of evidence weights spanning population constraint, variant consequence, inheritance compatibility, phenotype similarity, cardiac cell-state specificity, and prior gene–disease validity. The purpose is not to discover one numerically optimal recipe. It is to map where the top-ranked set changes and which evidence dimension causes the change. We record rank correlation, top-k overlap, evidence-layer agreement, and candidate-specific rank trajectories. Known cardiomyopathy genes serve as anchors rather than as a complete gold standard, because historical gene discovery is itself biased. Less-characterized genes are evaluated for coherent evidence profiles rather than rewarded for resemblance to every known gene. Failure modes tested A constraint-dominant model can favor essential developmental genes with little cardiac specificity. A phenotype-dominant model can favor heavily annotated genes. An expression-dominant model can favor abundant structural transcripts. We test each failure mode by ablating one layer, permuting weights, and comparing with matched controls. Candidates that remain near the top for incompatible reasons are flagged for manual review rather than treated as robust. We also separate syndromic and isolated cardiomyopathy representations. A gene with strong neurodevelopmental, metabolic, or multisystem phenotype concordance may be appropriate for a syndromic presentation even when a cardiac-only model ranks it lower. This prevents phenotype breadth from being mislabeled as noise. Technical findings Hard constraint thresholds produce unstable boundaries: candidates on either side of the cutoff can have similar underlying uncertainty, while highly constrained genes can remain elevated despite weak cardiac evidence. Continuous, uncertainty-aware constraint contributes useful separation without forcing a false categorical decision. The most stable priorities are supported by at least two evidence families beyond constraint—for example, compatible variant mechanism plus phenotype fit, or phenotype fit plus cardiomyocyte-state localization. Candidates supported almost entirely by one population metric move substantially under reasonable weight changes and are not ready to anchor an experimental program. Interpretive standard A rank is a navigation device. It does not convert a variant of uncertain significance into a causal allele, and it does not replace segregation, de novo assessment, allelic phase, phenotype review, or disease-specific variant criteria. The benchmark is designed to make those missing links visible. For every priority, we therefore report the evidence profile, the strongest competing explanation, the weight range over which the priority is stable, and the experiment or clinical-domain review that would most reduce uncertainty. Consolidated findings What the analysis establishes Continuous priors outperform hard gates conceptually Retaining the magnitude and uncertainty of constraint avoids artificial categorical boundaries. Mechanism matching is mandatory Loss-of-function and missense constraint answer different questions and must be aligned with the proposed allelic mechanism. Two-family support is a useful minimum Stable candidates require agreement between independent evidence families rather than repeated measurements of one prior. Rank trajectories reveal fragility How a candidate moves across reasonable weight choices is more informative than its position in one final list. Research conclusion Conclusion and experimental handoff We conclude that population constraint should be used as calibrated background evidence, never as an automatic proxy for disease relevance. In pediatric cardiomyopathy, the strongest research priorities are those for which allelic mechanism, inheritance, phenotype, and cardiac context remain coherent across plausible weighting schemes. The next step for a stable but less-characterized candidate is a mechanism-matched assay selected from the evidence profile: dosage perturbation for a credible haploinsufficiency model, variant-specific editing for an altered-protein model, and cardiac-lineage testing when cell-state evidence is central. An experiment that tests the wrong molecular direction cannot validate the computational ranking. Interpretive boundary Limitations Constraint estimates vary in power across genes and transcripts. Known disease genes are an incomplete and historically biased benchmark set. Public phenotype annotations may underrepresent age-dependent and incompletely penetrant features. Ranking stability does not establish variant causality or clinical actionability. Technical vocabulary Glossary LOEUF The upper confidence bound of the observed-to-expected ratio for predicted loss-of-function variants; lower values indicate stronger depletion. Prior ablation Removal of one evidence layer to measure how much it determines a result. Rank trajectory The movement of a candidate across a defined grid of model assumptions or evidence weights. Mechanism matching Alignment of the evidence metric and assay with the proposed molecular consequence. Reproducibility and evidentiary scope This study is a REELD public-data analysis and methods interpretation. It does not report a newly recruited clinical cohort, classify an individual variant, or replace clinical review. A release-ready execution of the workflow includes accession-level provenance, source and ontology versions, code state, environment locks, predefined sensitivity analyses, and output checksums. Continue exploring Browse all studies REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Cross-atlas replication of dosage-sensitive programs | REELD URL: https://reeld.org/studies/cross-atlas-dosage-programs Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All studies Replication study May 2026 Open analysis Cross-atlas replication of dosage-sensitive programs A study of which haploinsufficiency signals persist across public single-cell references. REELD research visualization. Image is illustrative and does not represent a measured result. Research question Which dosage-sensitive expression programs reproduce across independently generated cell atlases? Evidence layers Dosage-sensitivity references Multi-tissue single-cell atlases Hierarchical label harmonization Co-expression neighborhood analysis Matched gene-set resampling Leave-one-atlas-out testing Technical profile Biological unit Cell-contextualized gene program Primary contrast Cross-atlas transfer versus source-specific signal Inference level Program replication Decisive control Leave-one-atlas-out neighborhood preservation Primary analytical observation Replicable dosage-sensitive programs are defined by conserved biological neighborhoods and coherent cell contexts, not by a single marker, one atlas, or one label ontology. Gene-level constraint alone does not establish a shared cellular program. Executive interpretation Dosage-sensitive genes are often discussed as a single class, but they participate in very different biological systems. A constrained transcription factor, a ribosomal protein, a chromatin regulator, and a structural protein may all be depleted for predicted loss-of-function variation while acting in different cell types and developmental windows. This study asks which expression programs travel across independently produced atlases. We find that replication is strongest when defined at the level of a biological neighborhood—coherent sets of genes active in related cell contexts—rather than a universal dosage-sensitivity signature. Program definition Seed genes are drawn from curated dosage-sensitivity and population-constraint resources under explicit inclusion rules. Programs are then expanded using within-reference co-expression neighborhoods, but seed identity and neighbor identity remain distinct. This prevents a large inferred network from being mislabeled as direct dosage evidence. Neighborhoods are estimated within donor-aware cell strata and summarized across references. Highly ubiquitous housekeeping modules, mitochondrial read fractions, ribosomal abundance, and cell-cycle programs are modeled as competing explanations. Program size is capped and resampled so large modules do not obtain an automatic enrichment advantage. Cross-atlas harmonization Cell labels are represented hierarchically. Broad classes establish whether a program transfers across studies; fine states determine where the mechanism may be specific. Original labels, mapped labels, and mapping confidence are retained together. We do not force a one-to-one correspondence when atlases describe different biological resolution. Expression values are normalized within source, and replication is assessed through ranks, direction, neighborhood overlap, and cell-context concordance. Raw effect sizes are not directly pooled across incompatible platforms. A random-effects summary is used only when the underlying contrast and biological unit are sufficiently aligned. Replication criteria A program is considered transferable when its leading biological neighborhood remains enriched in the same broad lineage and compatible fine states across independent references, with no single atlas carrying the conclusion. Leave-one-atlas-out analysis must preserve the direction and the central functional interpretation. A marker that is reproducible without its neighbors is classified as a replicated gene signal, not a replicated program. Conversely, modest gene-level overlap can still support replication when orthologous or functionally equivalent neighborhoods preserve the same process. This distinction is essential in atlases with different feature detection and annotation depth. Technical findings Broad cellular processes associated with transcriptional regulation, proteostasis, translation, and developmental control frequently appear among dosage-sensitive genes, but their cell-context specificity varies. Some neighborhoods are widespread and likely reflect general cellular vulnerability; others localize to developmental or lineage-specific transitions and are more useful for disease mechanism design. The apparent strength of a dosage program often decreases after matching for baseline expression and annotation degree. The programs that remain coherent are carried by multiple genes and preserve a functional neighborhood across references. Atlas-specific programs are retained as hypotheses with narrow scope rather than discarded or generalized. Interpretation and use A replicated program can identify the cell system in which dosage perturbation is most likely to produce a measurable state change. It cannot by itself establish that reduced dosage is the disease mechanism for every seed gene. Variant consequence, inheritance, allelic series, and gene-specific functional evidence remain necessary. The practical output is a versioned program catalog with source provenance, neighborhood membership, transfer status, null-model performance, and explicit non-transfer conditions. That catalog is designed to support experimental selection and external reanalysis. Consolidated findings What the analysis establishes Neighborhoods replicate better than isolated markers Conserved functional neighborhoods are more portable across atlases than one highly expressed gene or one source label. Constraint does not imply one program Dosage-sensitive genes partition into distinct biological and cellular contexts. Label resolution changes specificity, not necessarily direction Broad lineage agreement can coexist with disagreement over fine-state nomenclature. Non-transfer is informative A program limited to one tissue, developmental period, or protocol defines a bounded hypothesis rather than a failed study. Research conclusion Conclusion and experimental handoff We conclude that dosage sensitivity becomes mechanistically useful when it is attached to a replicated cell-contextualized program. The most defensible programs are multi-gene, preserve their biological neighborhood, and survive removal of any one atlas. Experimental follow-up should perturb dosage in the cell context predicted by the transferable program and measure both the program-level response and gene-specific phenotypes. Rescue by restoring dosage provides substantially stronger evidence than reproduction of a generic stress response. Interpretive boundary Limitations Atlas coverage is uneven across tissues, developmental periods, and disease states. Co-expression is compatible with shared regulation but does not prove direct molecular interaction. Harmonized labels can obscure source-specific biology if mapping confidence is ignored. Dosage sensitivity inferred from population depletion is not equivalent to proven haploinsufficiency for a specific disorder. Technical vocabulary Glossary Dosage sensitivity A relationship in which altered gene copy number or expression level affects biological function or phenotype. Biological neighborhood A group of genes that repeatedly cohere by expression, pathway, or cell context. Random-effects summary A meta-analytic model that allows true effects to vary across studies. Transfer status A record of where a signal reproduces, partially reproduces, or fails to reproduce. Reproducibility and evidentiary scope This study is a REELD public-data analysis and methods interpretation. It does not report a newly recruited clinical cohort, classify an individual variant, or replace clinical review. A release-ready execution of the workflow includes accession-level provenance, source and ontology versions, code state, environment locks, predefined sensitivity analyses, and output checksums. Continue exploring Browse all studies REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Ontology depth changes gene-ranking stability | REELD URL: https://reeld.org/studies/ontology-depth-ranking-stability Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All studies Ontology study April 2026 Research brief Ontology depth changes gene-ranking stability Measuring how broad and specific phenotype terms alter computational prioritization. REELD research visualization. Image is illustrative and does not represent a measured result. Research question How much of a gene ranking is determined by the depth and propagation of phenotype terms? Evidence layers Human Phenotype Ontology Resnik and normalized semantic similarity Information-content analysis Leave-one-term-out ablation Ancestor-depth perturbation Rank concordance Technical profile Biological unit Phenotype representation Primary contrast Specific-term versus ancestor-expanded ranking Inference level Ranking stability Decisive control Leave-one-term-out influence analysis Primary analytical observation Specific terms provide discrimination, broad ancestors protect recall, and neither view is sufficient alone. The most reliable ranking is accompanied by a term-influence map showing which observations, ontology relationships, and propagation choices determine candidate order. Executive interpretation Phenotype-driven prioritization is often presented as though the phenotype list were a neutral input. It is not. The choice between a broad term and a specific descendant changes information content, candidate overlap, and the influence of missing or uncertain observations. Ontology version and propagation rules are therefore part of the model. We conclude that broad and specific representations should be reported together. Specific terms sharpen the differential when they are well supported; ancestor expansion protects against vocabulary mismatch and incomplete annotation. A ranking is trustworthy only when the analyst can identify which terms carry it and how it changes under plausible alternate representations. Phenotype representation Each observation is retained with term identifier, label, ontology release, onset, frequency, modifier, negation, and certainty when available. Exact terms are never replaced silently by ancestors. Instead, the analysis constructs parallel representations: exact-only, bounded ancestor expansion, full ancestor expansion, and information-content weighted expansion. Absent phenotypes are modeled separately from unobserved phenotypes. A documented absence can reduce support for a disease model when the feature is expected and ascertainment is reliable. Missing information is not negative evidence. Conflating the two produces false precision. Similarity models We compare most-informative-common-ancestor similarity with normalized measures that account for the information content of both terms. Pairwise term similarities are aggregated under best-match and directional schemes because a symmetric score can hide whether a candidate disease explains the observed phenotype or merely shares a few broad ancestors. Annotation frequency and ontology topology are release dependent. Information content is therefore computed from a named corpus, and both the ontology graph and annotation corpus are version locked. Recomputing one without the other changes the model even if the phenotype list is unchanged. Stability analysis We remove each phenotype term in turn, collapse it to successive ancestors, and perturb uncertain observations. Candidate ranks are compared with Spearman correlation, top-k overlap, reciprocal-rank change, and candidate-specific influence. These metrics expose both global stability and local swaps among the highest priorities. We also calculate redundancy between terms. Two observations may appear independent while sharing most of their ancestors and annotations. Without redundancy control, repeated description of one organ system can overwhelm a rarer but more discriminating feature from another system. Technical findings Specific terms generally increase separation among closely related disease models, but they also create fragility when based on uncertain interpretation or sparse annotations. Broad terms recover semantically adjacent candidates and reduce vocabulary mismatch, but they can flatten meaningful distinctions. The optimal representation is therefore conditional on observation quality and the decision being supported. The most concerning rankings are those dominated by one broad, highly connected term or by a cluster of redundant terms. Stable rankings distribute influence across multiple phenotype branches and retain a recognizable candidate core under bounded ancestor changes. Instability is not automatically a failure; it identifies which clinical observation or ontology relationship requires review. Reporting standard A phenotype-driven analysis should publish the exact term set, ontology release, annotation corpus, propagation rule, similarity function, aggregation function, treatment of negation, and term-influence analysis. A screenshot of a phenotype list is not a reconstructable model. We recommend presenting a dual ranking: a specificity-forward view for discrimination and a bounded-ancestor view for recall. Agreement defines a stable core; disagreement defines the review queue. Consolidated findings What the analysis establishes Specificity and recall are complementary Exact terms sharpen discrimination while ancestor expansion reduces vocabulary mismatch. Term influence should be visible Candidate order can be dominated by one observation, and that dependence must be reported. Missing is not absent Unrecorded phenotypes should not be used as negative evidence. Ontology release is a model version Graph and annotation changes can alter results even when the observed terms do not change. Research conclusion Conclusion and practical handoff We conclude that there is no responsible single phenotype ranking without a representation audit. The appropriate output is a stable candidate core, a sensitivity envelope, and a term-influence map that shows why candidates move. For downstream review, analysts should prioritize the observations whose clarification would most change the ranking. This converts ontology sensitivity into a practical phenotyping plan rather than treating it as computational noise. Interpretive boundary Limitations Disease annotations are incomplete and uneven across rare conditions. Information content depends on the selected annotation corpus and release. Clinical observation quality cannot be recovered computationally from an imprecise source record. Semantic similarity ranks explanatory proximity; it does not establish a molecular mechanism. Technical vocabulary Glossary Resnik similarity Similarity defined by the information content of the most informative common ancestor shared by two ontology terms. Best-match average An aggregation that pairs each term with its most similar counterpart before averaging. Term ablation Removal of one term at a time to quantify its effect on the result. Sensitivity envelope The range of plausible rankings produced by predefined reasonable modeling choices. Reproducibility and evidentiary scope This study is a REELD public-data analysis and methods interpretation. It does not report a newly recruited clinical cohort, classify an individual variant, or replace clinical review. A release-ready execution of the workflow includes accession-level provenance, source and ontology versions, code state, environment locks, predefined sensitivity analyses, and output checksums. Continue exploring Browse all studies REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Rare variants and microglial activation states | REELD URL: https://reeld.org/studies/microglial-activation-states Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All studies Mechanism study March 2026 Methods note Rare variants and microglial activation states Separating disease-associated state signatures from generalized inflammatory response. REELD research visualization. Image is illustrative and does not represent a measured result. Research question Can cross-dataset comparison distinguish specific microglial programs from common stress responses? Evidence layers Single-cell microglia references Rare-disease gene modules Signature decomposition Interferon and stress controls Donor-aware pseudobulk scoring Cross-dataset meta-analysis Technical profile Biological unit Microglial state program Primary contrast Disease-coherent residual versus generic activation Inference level Cell-state mechanism hypothesis Decisive control Negative-control signature decomposition Primary analytical observation Microglial enrichment is mechanistically informative only after generalized interferon, stress, cell-cycle, and dissociation components are modeled explicitly. The residual disease-coherent program, not the raw activation score, is the appropriate experimental hypothesis. Executive interpretation Activated microglia appear in many neurologic and systemic contexts. A strong enrichment can reflect a disease mechanism, a common interferon response, tissue dissociation, postmortem interval, cell stress, phagocytic activity, or shifts in cell composition. Calling all of these states “disease-associated” creates a biological category that is too broad to test. Our analysis treats activation as a mixture. We identify the component shared with common technical and inflammatory programs, subtract or condition on that component, and ask whether a residual rare-disease module reproduces across references. The conclusion is deliberately narrower but more actionable: only the reproducible residual should guide perturbation. Signature architecture Target modules are constructed from rare-disease genes and phenotype-linked pathways relevant to neuroimmune dysfunction. Control libraries include type I and type II interferon response, immediate-early response, heat shock, oxidative stress, cell cycle, hypoxia, apoptosis, ribosomal shifts, mitochondrial stress, and dissociation-associated programs. Each control is versioned and scored with the same procedure used for the target. Overlapping genes are not deleted automatically. Shared membership can be biologically meaningful. Instead, we decompose target activity into shared and residual components using regression, matched resampling, and leading-edge inspection. This shows whether the apparent signal is fully explained by a generic response or retains a coherent disease-linked component. Donor-aware analysis Microglial states can be represented by thousands of cells from few donors. We aggregate within donor and state, model donor as the independent unit, and avoid testing each cell as a replicate. When covariates are available, age, sex, tissue region, disease status, technical protocol, and postmortem characteristics are retained as design variables or stratification factors. State abundance and state expression are evaluated separately. More cells in a state can create an apparent program shift even when within-state expression is unchanged. Conversely, expression remodeling can occur without a major abundance change. Conflating the two obscures mechanism. Cross-dataset comparison References are compared at the level of broad homeostatic, inflammatory, phagocytic, interferon-responsive, proliferative, and stress-associated neighborhoods, while retaining source-specific labels. Directional meta-analysis is used when the underlying contrast is aligned; otherwise, transfer is assessed through rank and leading-edge concordance. We require the residual target program to preserve direction across independent datasets and to remain distinguishable from at least one expression-matched control set and the explicit negative-control library. A signal that transfers only in one preprocessing configuration is classified as unstable. Technical findings Raw microglial activation scores commonly share substantial structure with interferon, stress, phagolysosomal, and dissociation programs. Removing these components reduces the apparent breadth of the signal. That reduction is not a loss of discovery; it is the removal of explanations that are too general to support the proposed rare-disease mechanism. The residual programs that remain interpretable are smaller, state bounded, and carried by multiple genes rather than one canonical activation marker. They reproduce as coordinated neighborhoods more reliably than as identical cluster labels. Programs that disappear after control decomposition are reported as nonspecific activation, not as disease mechanisms. Falsifiable predictions A disease-coherent residual predicts that perturbing representative module genes in a defined microglial context will alter the residual program more strongly than matched generic stress controls. It also predicts that rescue or correction will reverse the residual without requiring global suppression of interferon or viability programs. The interpretation would be weakened if the residual is reproduced by unrelated stressors, disappears after donor-aware aggregation, is carried by one dataset, or fails to change under mechanism-matched perturbation. Consolidated findings What the analysis establishes Raw activation is not specific Common inflammatory and technical programs explain a meaningful fraction of apparent disease enrichment. Residualization sharpens the claim The remaining multi-gene component provides a narrower and more falsifiable hypothesis. Abundance and expression are different outcomes Changes in state frequency should not be conflated with within-state transcriptional remodeling. Donor is the inferential unit Cell counts increase measurement depth but do not create additional independent biological replicates. Research conclusion Conclusion and experimental handoff We conclude that a microglial rare-disease mechanism should be defined as a residual, replicated state program after generic activation and technical signatures have been accounted for. The raw activation label is too nonspecific to nominate a therapeutic direction. The experimental handoff is a factorial perturbation in microglia with disease-relevant and generic inflammatory stimuli, isogenic genetic perturbation, rescue, and parallel measurements of viability, cytokine response, phagocytic function, and the residual module. This design can distinguish a specific mechanism from generalized immune suppression. Interpretive boundary Limitations Postmortem and dissociation effects cannot be fully reconstructed from public metadata. Microglial states are spatially and temporally dynamic and may not transfer to monoculture. Residualization can remove true biology when a disease mechanism genuinely uses a common inflammatory pathway. Transcriptomic signatures do not establish protein activity, cell-cell signaling, or causal direction. Technical vocabulary Glossary Residual program The component of a target signature remaining after modeled generic or technical components are accounted for. Pseudoreplication Incorrect treatment of non-independent observations, such as cells from one donor, as independent replicates. State abundance The proportion or count of cells assigned to a state within a biological sample. Leading-edge inspection Review of the genes that carry a signature score to determine whether the same biology reproduces. Reproducibility and evidentiary scope This study is a REELD public-data analysis and methods interpretation. It does not report a newly recruited clinical cohort, classify an individual variant, or replace clinical review. A release-ready execution of the workflow includes accession-level provenance, source and ontology versions, code state, environment locks, predefined sensitivity analyses, and output checksums. Continue exploring Browse all studies REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # A reproducible evidence stack for inherited cardiomyopathy | REELD URL: https://reeld.org/case-studies/cardiomyopathy-evidence-stack Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Case studies Methods demonstration A reproducible evidence stack for inherited cardiomyopathy From heterogeneous variant annotations to a testable cardiomyocyte-state hypothesis. Research context The challenge A candidate list combined established sarcomeric genes, genes associated with syndromic cardiac disease, and less-characterized genes supported by uneven or partially circular evidence. A flat rank obscured why each candidate was present and which candidates remained plausible after the phenotype and allelic mechanism were specified. Analytical design The approach We rebuilt the list as a provenance-tracked evidence graph joining transcript-aware consequence, population frequency, inheritance, gene–disease validity, phenotype similarity, cardiac cell-state localization, and cross-atlas replication. Each layer was scored and stress-tested independently before integration. Decision value The outcome The integrated analysis separated a stable sarcomere-centered core from candidates dependent on one annotation source or one broad phenotype term. It produced a smaller experimental queue, named the appropriate cardiomyocyte context for each priority, and exposed the evidence still required before any variant-level claim. Case profile Decision Which candidates warrant mechanism-matched cardiac follow-up? Primary unit Variant–transcript–gene–phenotype chain Integration model Inspectable evidence graph Key control Layer ablation and cross-atlas replication Case framing at a high level The original candidate list had the familiar shape of a rare-disease analysis: a small number of genes with strong prior evidence, a longer tail of plausible candidates, and several entries whose rank could not be explained without reconstructing the pipeline. The practical question was not how to generate another score. It was how to identify the candidates for which the genomic, phenotypic, and cellular arguments genuinely agreed. We defined the decision in advance: select a bounded set of research priorities and specify the cardiac experiment that would most efficiently challenge each one. Clinical classification was explicitly outside scope. That boundary allowed us to treat uncertainty as an analytical object rather than pressure the evidence into a diagnostic category. Reconstructing the evidence graph Variant records were normalized to a consistent genome build and transcript set. Consequence disagreements, low-confidence predicted loss-of-function annotations, splice predictions, population frequencies, and domain positions were retained as separate evidence nodes. Gene-level constraint was connected to, but not merged with, variant-level rarity. Disease validity, inheritance, and phenotype annotations were versioned independently. The graph prevented double counting. A curated disease association and a phenotype similarity score can derive from the same underlying literature; treating them as independent confirmation inflates confidence. We tagged shared provenance and grouped evidence by family so repeated representations of one fact could not masquerade as convergence. Variant consequence and transcript relevance. Population frequency and constraint with uncertainty. Inheritance, segregation availability, and allelic mechanism. Phenotype fit across cardiac and extracardiac branches. Cardiomyocyte state and pathway localization. Independent-reference transfer and competing explanations. Phenotype and mechanism stratification The phenotype representation was divided into structural myocardial disease, rhythm and conduction features, functional impairment, developmental or congenital findings, skeletal-muscle involvement, and metabolic or multisystem features. Broad cardiac ancestors were used for recall, while specific observations controlled discrimination. Term influence was inspected so one general label could not determine the ranking. Candidates were then stratified by proposed allelic mechanism. A dominant altered-protein hypothesis, a haploinsufficiency hypothesis, a recessive loss-of-function model, and a mitochondrial or metabolic mechanism require different priors and different experiments. This step materially changed the interpretation because the same gene-level evidence can support one mechanism while contradicting another. Cardiac cell-state integration Cardiac single-cell references were analyzed with donor-aware summaries and hierarchical cell labels. We tested whether candidate evidence localized beyond broad heart expression to ventricular or atrial cardiomyocyte programs, conduction-associated neighborhoods, fibroblast remodeling, vascular compartments, or immune states. Expression-matched controls addressed the tendency of abundant cardiac genes to dominate. State localization was interpreted as an experimental-context prior. It did not convert a candidate into a disease gene. A candidate supported by a contractile phenotype and ventricular sarcomere program suggested a cardiomyocyte mechanics assay; a candidate supported by a multisystem metabolic phenotype and broad energetic program suggested a different model and readout. Stress tests and decision rule We recomputed priorities after removing constraint, phenotype, prior disease knowledge, and cell-state evidence one layer at a time. We varied evidence weights, ontology depth, transcript choice, and atlas inclusion. Candidate-specific rank trajectories showed whether a priority was supported broadly or sat at the top only under one convenient configuration. The final research queue required support from at least two independent evidence families, mechanism compatibility, and a named disconfirming observation. Candidates that were biologically interesting but fragile were not discarded; they were moved to a clarification queue with the exact missing evidence identified. What the analysis changed The stable core was enriched for genes whose known or proposed biology aligned with sarcomere organization, cardiomyocyte mechanics, and the represented cardiac phenotype. Some less-characterized candidates remained credible because their cellular and phenotypic evidence converged independently of constraint. Others fell when a broad cardiomyopathy term, a single annotation database, or general cardiac abundance was removed. The most important change was not the ordering of the list. It was the conversion of each retained candidate into a testable statement: proposed molecular direction, relevant cardiac state, measurable cellular consequence, and result that would weaken the hypothesis. Case findings What the analysis establishes A stable sarcomere-centered core emerged Candidates with compatible allelic mechanism, phenotype, and ventricular contractile context remained prioritized across model perturbations. Several high ranks were prior dependent Some candidates moved sharply when gene constraint or one disease annotation source was removed. Cell context changed the experiment The same gene rank implied different follow-up depending on whether support localized to cardiomyocyte mechanics, remodeling, conduction, or multisystem metabolism. Uncertainty became actionable Fragile candidates were assigned specific clarification needs instead of being labeled generically uncertain. Case conclusion What we conclude We conclude that a reproducible evidence stack is more valuable than a definitive-looking composite score. For inherited cardiomyopathy, the strongest research priorities are the candidates whose variant mechanism, phenotype branch, and cardiomyocyte program remain coherent after removal of any one evidence source. The immediate experimental priorities are mechanism matched: isogenic editing and rescue for variant-specific altered-protein hypotheses; dosage titration for credible haploinsufficiency; engineered cardiac tissue or force measurements for sarcomere-centered mechanisms; and orthogonal metabolic assays for multisystem energetic hypotheses. A candidate should leave the research queue only when the proposed direction and context survive direct perturbation. Interpretive boundary Limitations The demonstration uses public evidence patterns rather than a prospectively enrolled clinical cohort. Segregation, phase, penetrance, and individual-level phenotyping are not recoverable from aggregate public resources. Single-cell expression provides context but does not measure variant-specific protein function. Well-studied genes have denser evidence graphs, requiring explicit correction for annotation advantage. Interpretive boundary This case study is a computational methods demonstration. It does not establish pathogenicity, diagnose an individual, estimate penetrance, or replace disease-specific variant interpretation and clinical review. Bring us a hard mechanism question. We collaborate on public-data analyses that need stronger context, controls, or translational logic. Collaborate REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Finding a developmental window in epilepsy genetics | REELD URL: https://reeld.org/case-studies/developmental-epilepsy-cell-window Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Case studies Methods demonstration Finding a developmental window in epilepsy genetics Using cortical atlases to locate when a phenotype-linked gene module is most coherent. Research context The challenge Genes associated with developmental epilepsy appeared broadly expressed across neural tissue. Tissue-level summaries therefore confirmed brain relevance without identifying a developmental window, lineage, or measurable transition for perturbation. Analytical design The approach We decomposed the phenotype into seizure, developmental, behavioral, motor, and morphologic branches; built evidence-weighted gene modules; and evaluated them across developmental cortical trajectories with donor-aware scoring, matched null sets, and leave-one-atlas-out transfer. Decision value The outcome A bounded subset of the module remained coherent during excitatory-neuron differentiation and maturation, while broad neuronal and stress components were removed. The result specified when to perturb, what state transition to measure, and which findings would contradict a shared-window hypothesis. Case profile Decision Which developmental stage should carry the perturbation? Primary unit Phenotype-linked gene module Integration model Trajectory-aware cross-atlas transfer Key control Matched module resampling and term ablation Case framing at a high level A list of epilepsy-associated genes can be strongly brain expressed without revealing where their effects converge. Neuronal genes are often detected across multiple cortical populations, and disease annotations frequently share broad terms. The initial analysis therefore produced a biologically plausible but experimentally vague conclusion: the module was neuronal. We reframed the decision around developmental timing. The question became whether a reproducible subset of genes aligned with a transition that could be modeled—progenitor proliferation, neuronal specification, migration, excitatory-neuron maturation, inhibitory-neuron development, or synaptic network assembly. Phenotype decomposition Seizure type and onset were separated from global developmental delay, intellectual disability, regression, tone abnormalities, movement findings, sleep disturbance, behavioral features, and structural brain abnormalities. Exact and ancestor-expanded representations were analyzed in parallel. Leave-one-term-out analysis identified which clinical observations carried the module definition. This reduced circularity. Genes were not considered convergent merely because all were already labeled with a broad epilepsy term. The analysis required the phenotype-defined module to show additional developmental structure in independent expression data. Trajectory-aware cellular analysis Developmental cortical atlases were mapped to conservative lineage stages while preserving original annotations. Module activity was summarized per donor and state. The analysis compared broad lineages for transfer and fine states for mechanism specificity. Pseudotime was treated as an inferred ordering, not literal chronological age. Matched null sets controlled for gene length, baseline expression, annotation degree, and generic neuronal membership. Cell-cycle, immediate-early, ribosomal, and general synaptic programs were scored alongside the target module. A developmental window was retained only if the target carried information beyond these controls. Cross-atlas validation The primary result was repeated after excluding each atlas, donor group, and phenotype branch. We tracked direction, peak trajectory position, leading-edge genes, and neighborhood overlap. Exact cluster names differed across references, so replication was defined by compatible developmental ordering and cell lineage rather than lexical identity. A subset of candidate genes shifted out of the stable leading edge when one broad developmental term was removed or when expression matching was applied. Those genes remained relevant to epilepsy biology but did not support the narrower shared-window conclusion. Mechanistic interpretation The stable module was most coherent after excitatory-neuron identity emerged and during maturation of synaptic, ion-channel, and activity-regulated programs. This does not imply one molecular pathway. It indicates that distinct upstream perturbations may become phenotypically legible during a shared cellular transition. The analysis did not support a universal neuronal vulnerability claim. Inhibitory-neuron, progenitor, and glial programs remained relevant for subsets of genes, but they did not carry the same integrated module after resampling and phenotype refinement. The correct conclusion is modular convergence with a defined boundary. Experimental decision The chosen handoff is a staged perturbation design in which representative genes are edited before and after the predicted transition. Readouts include progression through state markers, neurite and synapse morphology, intrinsic excitability, network activity, and recovery under gene correction or mechanism-specific rescue. The shared-window model predicts stronger and more coherent effects near the nominated maturation stage than in an undifferentiated progenitor state. Failure of that timing prediction would require revision even if a general cellular phenotype remained. Case findings What the analysis establishes Brain expression was not the finding Broad neuronal detection confirmed tissue relevance but did not identify a mechanistic window. A maturation-stage module survived controls A bounded leading edge remained coherent during excitatory-neuron differentiation and maturation across references. The convergence was partial Progenitor, inhibitory-neuron, and glial contexts remained plausible for subsets and were not forced into one model. Timing became a falsifiable variable The analysis generated a prediction about when perturbation should have its strongest state-level effect. Case conclusion What we conclude We conclude that the most useful shared context for this phenotype-linked module is a defined excitatory-neuron maturation window, not the brain or neurons in general. The conclusion is supported by trajectory position, leading-edge stability, and cross-atlas direction after matched controls. A valid follow-up must test timing. Perturbations introduced at multiple developmental stages, with rescue and orthogonal functional readouts, can determine whether the shared window represents mechanism or merely correlated expression. Gene-specific deviations should be preserved rather than interpreted as experimental failure. Interpretive boundary Limitations Cortical atlases sample different regions, gestational stages, and protocols. Inferred trajectories are not direct lineage tracing and do not represent exact developmental time. Epilepsy phenotypes and gene annotations are heterogeneous and unevenly curated. Organoid and induced-neuron models may incompletely reproduce in-vivo maturation and circuit context. Interpretive boundary Cell-state coherence identifies an experimentally informative context. It does not prove that every gene in the module shares one molecular mechanism or that an in-vitro trajectory reproduces human cortical development. Bring us a hard mechanism question. We collaborate on public-data analyses that need stronger context, controls, or translational logic. Collaborate REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Negative controls sharpen an immune-dysregulation signal | REELD URL: https://reeld.org/case-studies/immune-dysregulation-negative-controls Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Case studies Methods demonstration Negative controls sharpen an immune-dysregulation signal Testing a rare-disease module against common inflammatory and technical signatures. Research context The challenge An apparent monocyte enrichment could not be distinguished from generic interferon signaling, acute stress, cell-cycle shifts, and sample-processing effects. The initial result was strong in magnitude but weak in specificity. Analytical design The approach We constructed a versioned negative-control library, analyzed donor-level pseudobulk profiles, separated state abundance from within-state expression, and decomposed the target module into generic and residual components across independent public immune references. Decision value The outcome Most of the broad activation signal was explained by common inflammatory programs, but a smaller multi-gene residual reproduced in a bounded monocyte state. That residual—not the raw enrichment—became the candidate disease mechanism and the basis of a factorial perturbation plan. Case profile Decision Is the monocyte signal disease coherent or generically activated? Primary unit Donor-by-cell-state expression Integration model Control-signature decomposition Key control Interferon, stress, cycle, and dissociation library Case framing at a high level The first analysis produced an intuitively attractive story: rare-disease genes were enriched in activated monocytes, suggesting an innate-immune mechanism. The problem was that the same state also carried interferon, stress, and sample-processing signatures observed across unrelated conditions. Magnitude alone could not distinguish specificity. We changed the question from “is the module enriched?” to “what part of the module cannot be explained by common activation?” This made negative controls a primary analytical layer instead of a supplementary figure. Negative-control library The control library included type I and type II interferon response, NF-kB-associated inflammatory activation, immediate-early response, heat shock, oxidative stress, hypoxia, apoptosis, cell cycle, ribosomal and mitochondrial quality shifts, and dissociation-associated programs. Each signature retained its source, context, gene identifiers, direction, and known limitations. Controls were scored with the same method and background used for the target module. This symmetry matters: a target tested with a favorable scoring method and controls tested differently do not provide a valid specificity comparison. Overlap was decomposed rather than removed indiscriminately because common inflammatory pathways can be part of genuine disease biology. Donor and state model Counts were aggregated within donor-by-state strata, and donor remained the unit of inference. Cell number entered as a precision and quality variable, not as the sample size. State abundance and within-state differential expression were estimated separately to distinguish recruitment or expansion from transcriptional remodeling. Where metadata permitted, age, sex, disease context, stimulation, processing batch, and collection protocol were included as design variables or used for stratified sensitivity analyses. Analyses with unresolved confounding were labeled as transportability tests rather than causal contrasts. Signal decomposition The target score was modeled against the control-signature matrix, and the residual was inspected for multi-gene coherence. Regression residuals, matched resampling, and leading-edge overlap provided complementary views. No single residualization method was allowed to define the conclusion. The raw monocyte enrichment decreased substantially after generic activation components were accounted for. The remaining signal was smaller but more stable across references and less correlated with global RNA quality, cell cycle, or interferon intensity. Several canonical activation genes left the leading edge, while a bounded set of target-linked genes remained coordinated. Specificity and alternative explanations The residual was compared with expression-matched random modules and tested in unrelated inflammatory contexts. It did not behave as a universal marker of all activated monocytes, but neither was it exclusive to one rare-disease label. We therefore interpret it as a candidate mechanistic state shared by a narrower set of immune perturbations. Medication, infection, tissue compartment, and unrecorded processing remain plausible alternatives. The analysis cannot eliminate them from public data. It can show that the conclusion no longer depends entirely on the most common technical and inflammatory programs. Experimental decision The resulting experiment uses a factorial design: genotype or gene perturbation crossed with disease-relevant stimulation and generic interferon or stress controls. The readout panel includes the residual module, cytokine secretion, viability, differentiation state, and a function linked to the nominated mechanism. A mechanism-specific result should alter the residual program and functional readout without simply reproducing global stress or suppressing all inflammatory signaling. Genetic correction or pathway-specific rescue should reverse the residual component. If only generic activation changes, the rare-disease interpretation should be rejected or narrowed. Case findings What the analysis establishes The original signal was overbroad Common interferon, stress, and processing programs explained much of the raw enrichment. A smaller residual reproduced A multi-gene component remained coordinated in a bounded monocyte state across independent references. Specificity improved as magnitude decreased The reduction in score represented removal of generic explanations, not loss of the central hypothesis. The experiment changed materially The final design tests disease-relevant and generic stimuli side by side and requires rescue of the residual. Case conclusion What we conclude We conclude that the raw activated-monocyte signal is not sufficiently specific to support a rare-disease mechanism. After explicit decomposition, a smaller residual program remains credible as a research hypothesis because it is multi-gene, donor-aware, and transferable across references. The residual must now earn causal status experimentally. A factorial perturbation with matched stress and interferon controls, functional readouts, and rescue can determine whether it represents a disease-linked immune state or another broadly inducible response. Interpretive boundary Limitations Public immune references contain incomplete covariate and treatment metadata. Residualization can remove true biology when common inflammatory pathways are mechanistically central. Peripheral immune states may not reproduce tissue-resident or disease-site biology. The case demonstrates evidence refinement, not clinical subgroup discovery. Interpretive boundary Public references cannot resolve every cohort-specific covariate, medication effect, infection history, or handling artifact. The residual program remains a research hypothesis requiring primary-data and experimental validation. Bring us a hard mechanism question. We collaborate on public-data analyses that need stronger context, controls, or translational logic. Collaborate REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Reconciling multisystem phenotypes in mitochondrial disease | REELD URL: https://reeld.org/case-studies/mitochondrial-phenotype-reconciliation Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Case studies Methods demonstration Reconciling multisystem phenotypes in mitochondrial disease Using ontology structure to preserve organ-specific evidence without fragmenting the case. Research context The challenge Neurologic, cardiac, muscular, ophthalmologic, and metabolic observations produced competing gene rankings when analyzed as independent lists. Pooling all terms improved apparent coherence but allowed the largest phenotype branch to dominate. Analytical design The approach We represented the phenotype as a hierarchical, organ-aware graph; modeled exact terms, informative ancestors, onset, frequency, and negation; and compared branch-balanced semantic similarity with conventional pooled rankings under term ablation and ontology-version sensitivity. Decision value The outcome The analysis identified a stable cross-system core supported across multiple phenotype branches and a set of organ-dependent alternatives. It also showed which additional phenotyping or biochemical evidence would most reduce the remaining ambiguity. Case profile Decision Which candidates explain the multisystem pattern as one mechanism? Primary unit Organ-aware phenotype graph Integration model Branch-balanced semantic similarity Key control Organ-branch and term ablation Case framing at a high level Multisystem disease creates a representational problem. If neurologic, cardiac, muscle, vision, hearing, and metabolic features are analyzed separately, the result fragments into competing partial explanations. If they are pooled without structure, the branch with the most terms can dominate even when many terms are redundant. We defined the desired conclusion as a cross-system explanation that remains plausible when any one organ branch is reduced or removed. Organ-specific alternatives were retained rather than penalized, because real cases can involve blended phenotypes, secondary effects, incomplete ascertainment, or more than one process. Organ-aware phenotype graph Observed terms were organized into neurologic, cardiac, skeletal-muscle, ophthalmologic, auditory, endocrine, hepatic, renal, and biochemical branches. Onset, progression, episodic features, triggers, frequency, and documented absence were retained. Exact terms and informative ancestors remained separately addressable. Within each branch, redundant terms were down-weighted so multiple descriptions of one feature did not count as independent evidence. Across branches, scores were balanced to prevent the most densely phenotyped system from overwhelming the analysis. Weighting schemes were predefined and included equal-branch, observation-quality, and information-content views. Candidate and mechanism layers Candidate relationships included nuclear and mitochondrial genomic mechanisms, inheritance, subcellular process, tissue expression, and disease annotations. Oxidative phosphorylation, mitochondrial translation, dynamics, quality control, substrate metabolism, and cofactor pathways were kept distinct. “Mitochondrial” was not treated as one pathway. Variant consequence and inheritance compatibility were evaluated independently of phenotype similarity. Tissue-specific heteroplasmy, threshold effects, and maternal inheritance cannot be inferred from an ontology score, so mitochondrial-DNA candidates retained explicit missing-data flags. Sensitivity and reconciliation We removed each organ branch, each high-influence term, and each biochemical observation in turn. We compared exact-term, ancestor-expanded, branch-balanced, and conventionally pooled rankings. Candidate stability was summarized as a branch-support profile rather than one aggregate score. The pooled model over-weighted the most extensively documented branch. Branch balancing recovered candidates supported by multiple systems even when no single branch produced the top score. It also exposed alternatives supported almost entirely by one organ system, which are important but represent a different explanatory claim. Technical findings A stable core remained when neurologic, cardiac, and muscle branches were perturbed individually. These candidates were supported by shared energetic or organelle-maintenance biology and by nonredundant phenotype evidence across systems. The core was not identical to the top of any one organ-specific list. Several alternatives were strongly dependent on a single branch or one highly specific observation. Instead of discarding them, we linked each to the confirmatory evidence most likely to change the decision: focused biochemical testing, imaging review, tissue-specific molecular analysis, inheritance clarification, or more precise onset and progression data. Decision value The reconciled output separates three categories: cross-system core candidates, organ-dependent alternatives, and representation-sensitive candidates. Each category implies a different next step. The first supports integrated mechanism review, the second supports targeted specialty or biochemical evaluation, and the third supports phenotype clarification before more computation. This structure is more honest than a single list and more useful than separate organ lists. It shows which candidate explains the pattern, which explains only part of it, and which appears high because of how the phenotype was encoded. Case findings What the analysis establishes A cross-system core survived branch ablation Several candidates retained support across neurologic, cardiac, and muscle representations rather than depending on one organ list. Pooled rankings were branch-size sensitive The system with the most recorded terms exerted disproportionate influence without redundancy and branch balancing. Organ-dependent alternatives remained valuable Strong single-branch candidates were retained with targeted confirmation needs instead of forced into a unified mechanism. The model prioritized new observations Term influence identified the phenotyping and biochemical questions most likely to resolve candidate order. Case conclusion What we conclude We conclude that multisystem phenotype reconciliation requires a branch-aware model. The strongest cross-system candidates are those supported by nonredundant observations across organs and stable to removal of any one phenotype branch. A pooled term list cannot provide that assurance. The next step is evidence acquisition, not automatic selection of the highest rank. Biochemical assays, inheritance review, tissue-aware molecular analysis, and clarification of high-influence phenotypes should be chosen according to the candidate’s branch-support profile. This makes the computational result a plan for resolving uncertainty. Interpretive boundary Limitations Ontology annotations incompletely represent severity, longitudinal course, and quantitative biochemical measurements. Tissue-specific heteroplasmy and threshold effects are not inferable from aggregate phenotype similarity. Branch balancing involves value judgments that must be reported and stress-tested. A stable cross-system fit can still reflect correlated secondary effects rather than one primary mechanism. Interpretive boundary This is a computational evidence synthesis. It does not establish a mitochondrial diagnosis, determine heteroplasmy, infer tissue-specific variant load, or replace biochemical, molecular, and clinical assessment. Bring us a hard mechanism question. We collaborate on public-data analyses that need stronger context, controls, or translational logic. Collaborate REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Reproducibility is a threshold, not a slogan | REELD URL: https://reeld.org/editorials/reproducibility-is-a-threshold Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All editorials REELD Editorial August 19, 2026 18 min read Reproducibility is a threshold, not a slogan A mechanism that disappears under a reasonable alternate pipeline is not ready to anchor a therapeutic hypothesis. Editorial scope This is a REELD institutional perspective intended to support rigorous scientific review. It has not undergone external journal peer review. Review questions What is the biological unit of replication? Which analytical choices could plausibly change the conclusion? Was the validation reference selected before the primary result was interpreted? Which dimension of the result actually replicated? What outcome would force a narrower or opposite conclusion? Our position Computational biology can produce a coherent narrative from nearly any sufficiently large dataset. Dimensionality reduction will separate cells, enrichment tools will return pathways, and network methods will connect genes. None of those outputs is automatically false, but none is a mechanism merely because it is visually persuasive. REELD treats reproducibility as an admission threshold for mechanistic interpretation. Before a result can anchor a therapeutic hypothesis, it should survive reasonable changes in data source, preprocessing, label resolution, background universe, ontology representation, and statistical model. The required tests depend on the claim, but the obligation to define them does not. Reproduction, robustness, and replication are different Computational reproduction asks whether the same code and inputs generate the same output. Analytical robustness asks whether the conclusion persists under reasonable alternate choices. Biological replication asks whether compatible evidence appears in independent samples or systems. A study can pass the first and fail the other two. These distinctions should appear in the manuscript and release record. Containerizing an unstable analysis makes it repeatably unstable. Conversely, an independent dataset can support the same biological interpretation even when its platform and labels prevent exact numerical reproduction. The standard should match the inferential level of the claim. Reproduction: same inputs, code, environment, and deterministic output. Robustness: stable interpretation across predefined reasonable analytical choices. Replication: compatible evidence in independent biological data or experimental systems. The multiverse should be designed before the preferred result Every analysis contains a multiverse of plausible decisions: filtering thresholds, normalization, covariates, cell labels, gene universes, similarity functions, integration methods, and exclusion rules. If those decisions are explored after the result is visible, the analyst can unintentionally optimize for coherence. A convincing final figure may be one selected member of a much less stable result family. We favor an analysis charter that names the primary model, the reasonable alternatives, the negative controls, and the failure conditions before final interpretation. Not every possible pipeline deserves equal weight. The purpose is to define the scientifically defensible neighborhood around the primary analysis and measure whether the conclusion lives throughout it. Stability is multidimensional A result need not reproduce with identical effect size to remain useful. Direction, rank, cell-context neighborhood, leading-edge genes, pathway interpretation, and experimental implication can each be stable or unstable. Reporting one p-value or overlap coefficient compresses these dimensions and can hide a mechanistically important disagreement. For rare-disease mechanism work, we ask at least four questions. Does the candidate remain supported? Does the relevant cell or developmental context remain compatible? Does the molecular direction remain the same? Does the proposed experiment remain appropriate? A signal that preserves rank but reverses direction has not replicated in the way translation requires. Negative results define the transport boundary Failed transfer is not an embarrassment to be moved into supplementary material. It defines where the mechanism does not travel. A program present in fetal cortex but absent from adult references may be developmentally restricted. A signal present after dissociation but absent from nuclei may be technically contingent. A phenotype ranking that changes after one term is removed may depend on ascertainment quality. The correct response is not always rejection. It may be a narrower claim. Scientific value increases when the boundary is explicit because another group can choose the right tissue, stage, assay, and comparison rather than attempting to generalize the result blindly. What reviewers should demand Review should move beyond asking whether code is available. Reviewers should be able to reconstruct source accessions and releases, understand the unit of inference, inspect biological-replicate handling, evaluate the background model, see sensitivity to key choices, and identify the observation that would weaken the conclusion. Availability without auditability is insufficient. A result that fails these tests may still be an exploratory observation. It should be labeled accordingly. The problem is not exploration; it is presenting exploration with the rhetorical certainty of confirmation. What replication cannot establish Agreement across public datasets does not establish causality, clinical validity, target safety, or therapeutic efficacy. It reduces the space of fragile explanations and identifies a more precise experiment. That narrower claim is both defensible and useful. The final bridge remains perturbational. A computational mechanism earns stronger status when a mechanism-matched intervention changes the predicted molecular and cellular readouts, and when correction or rescue reverses the effect. Replication improves the target; it does not replace the test. Editorial standard Positions we apply in review A deterministic rerun is necessary but not sufficient for reproducibility. The biological replicate—not the cell, read, or image—is usually the relevant inferential unit. Reasonable alternate analyses should be specified before final interpretation. Instability and failed transfer belong in the primary scientific record. Computational replication narrows a hypothesis; perturbation tests causality. REELD conclusion Our conclusion Our editorial conclusion is direct: a mechanism that exists only in one convenient pipeline is not ready to guide translation. REELD will treat sensitivity, non-transfer, and disagreement as results. We would rather publish a narrower mechanism with a known boundary than a comprehensive story that cannot survive independent scrutiny. Continue exploring Read more editorials REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # A phenotype ontology is part of the model | REELD URL: https://reeld.org/editorials/phenotype-ontology-is-a-model Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All editorials Methods Editorial July 22, 2026 16 min read A phenotype ontology is part of the model Terms, ancestors, and information content shape the answer. They belong in the methods, not in a footnote. Editorial scope This is a REELD institutional perspective intended to support rigorous scientific review. It has not undergone external journal peer review. Review questions What source observation justified each selected term? Which release and annotation corpus were used? How were ancestors, negation, onset, and missingness handled? Which terms dominate the highest-ranked candidates? Does the conclusion survive a bounded alternate representation? Our position A phenotype ontology is not a neutral thesaurus placed between observation and algorithm. It is a versioned graph that encodes decisions about categories, relationships, granularity, and knowledge. Once a phenotype list enters semantic similarity, gene prioritization, cohort matching, or disease classification, the ontology is part of the computational model. That means ontology release, annotation corpus, term selection, propagation, negation, onset, frequency, and similarity function belong in the methods. A ranking without this information cannot be reconstructed and should not be presented as though it followed directly from the patient or disease description. Observation and representation must remain distinct The observation is what was assessed and found. The representation is the ontology term chosen to encode it. Those are related but not identical. Two observers can map the same clinical description to different term depths, and two ontology releases can place the same term in a different neighborhood. A responsible record retains source language, selected term, mapping confidence, modifiers, and alternatives when ambiguity matters. Silent replacement of a specific observation with a broad ancestor may improve recall while erasing the distinction required for mechanism. Silent over-specification can create precision the source never supported. Ancestors improve recall and create dependence Ontology propagation allows a specific observation to support comparisons at broader conceptual levels. This is useful when two records use different granularity. It also creates correlated features: a single observed term may contribute itself and many ancestors. Treating every propagated node as independent evidence multiplies one observation. Propagation should therefore be bounded and explicit. Information-content weighting can reduce the influence of common ancestors, but information content is itself corpus dependent. There is no universal specificity value detached from the annotations used to estimate it. Missing, absent, uncertain, and unassessed are not synonyms A documented absent phenotype can be informative when the feature is expected, age appropriate, and adequately assessed. A missing field carries no such meaning. An uncertain observation should not be forced into either category. Algorithms that collapse these states produce confident but biologically misleading differentials. The same applies to onset and frequency. A feature expected only later in life cannot be used as strong negative evidence in a young individual. A rare feature can be highly discriminating when present but weak evidence against a disease when absent. The phenotype model must preserve this asymmetry. Similarity scores hide direction A symmetric similarity score can conceal whether a disease model explains the observed features or whether the two records merely share broad ancestors. Directional coverage—how much of the observed phenotype a candidate explains and how much of the candidate phenotype was assessed—can reveal this difference. Composite scores also hide term influence. Two candidates can receive similar scores for completely different reasons. Review requires a term-by-candidate explanation showing the most informative matches, unmatched observations, documented contradictions, and sensitivity to term removal. Release drift is scientific drift Ontologies add terms, revise definitions, and change parent-child relationships. Annotation files add diseases, update frequencies, and alter evidence. A result can therefore change without a new observation or a code change. If the release is not locked, the model has changed silently. We recommend periodic release-delta analysis for active projects. When rankings change, the responsible graph or annotation update should be identified. Knowledge-base evolution is not noise; it is new model input and should be treated with the same provenance discipline as a new genomic reference. A minimum reporting standard Publish the exact term set, source descriptions, mapping confidence, ontology release, annotation corpus, propagation depth, similarity metric, aggregation rule, treatment of missingness and negation, and leave-one-term-out analysis. If a ranking changes materially under a reasonable alternate representation, publish both views. The correct endpoint is not the appearance of stability. It is knowledge of the stability envelope and the clinical observations most likely to reduce it. Editorial standard Positions we apply in review Phenotype encoding is a modeling step and must be auditable. Exact observations and propagated ancestors should remain distinguishable. Missing information must not be treated as an absent phenotype. Information content is corpus and release dependent. Every phenotype ranking should include term-influence and release-sensitivity analysis. REELD conclusion Our conclusion Our editorial conclusion is that phenotype ontology choices are substantive scientific assumptions. A ranking without its representation model is not portable evidence. REELD will publish the phenotype graph, not merely the final list, and will use sensitivity to identify where better observation is more valuable than more computation. Continue exploring Read more editorials REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Public data deserves primary-evidence discipline | REELD URL: https://reeld.org/editorials/public-data-is-primary-evidence Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All editorials Data Stewardship Editorial June 14, 2026 19 min read Public data deserves primary-evidence discipline Secondary analysis is not second-class science, but it must respect cohort design, provenance, and consent boundaries. Editorial scope This is a REELD institutional perspective intended to support rigorous scientific review. It has not undergone external journal peer review. Review questions Can every derived object be traced to a source release and transformation? Does the statistical model preserve donor and cohort structure? Were access terms and participant context respected? Which source differences could explain the result? Can an authorized external group reconstruct the analysis? Our position Public genomic, single-cell, and phenotype resources make questions testable that no single laboratory could address alone. Their scale creates scientific leverage. It does not remove the original study design, participant context, measurement error, access conditions, or consent boundary. Secondary analysis should therefore be conducted with primary-evidence discipline. The source cohort, assay, tissue acquisition, processing, reference versions, exclusion rules, and permitted uses remain part of every derived result. Public availability is a distribution condition, not a quality certificate or blanket ethical permission. The source is part of the result A cell atlas is not simply a matrix. It is a collection of donors, tissues, protocols, capture efficiencies, quality thresholds, and author-defined labels. A variant database is not simply an allele count. It has ascertainment, ancestry composition, coverage, calling, filtering, and annotation assumptions. A phenotype file is not simply a graph. It reflects curation sources, evidence codes, frequency, and release history. Pooling sources without retaining these structures creates synthetic certainty. The combined sample size increases while the ability to explain what was measured decreases. We prefer federated summaries or study-aware models when direct pooling would erase the unit of evidence. Provenance before harmonization Every source should enter with an accession, release, download date, license or access terms, file identity, cohort description, sample exclusions, and source citation. Every transformation should record identifier mapping, normalization, filtering, label changes, reference builds, and software environment. Derived objects should be traceable to inputs and code state through immutable identifiers or checksums. Harmonization must preserve disagreement. If two sources assign different cell labels or variant consequences, the mapping should carry confidence and alternatives. A unified table that discards uncertainty is easier to analyze and harder to trust. Cohort and donor structure cannot be repaired by cell count Single-cell resources can contain millions of cells from a modest number of donors. The cells increase resolution; they do not create millions of independent biological replicates. Analyses that ignore donor structure underestimate variance and can produce false confidence. Similar issues arise when multiple tissues, timepoints, or family members are treated as independent without acknowledging hierarchy. Metadata limitations should change the claim. If critical covariates are absent, an analysis may support transportability or hypothesis generation but not a causal contrast. Calling the same model “adjusted” because a batch variable was available does not solve unmeasured confounding. Open access does not erase participant context Data can be legally accessible and still require ethical restraint. Researchers should use the minimum resolution necessary, avoid re-identification attempts, honor controlled-access and downstream-use limitations, and consider whether group-level interpretations could stigmatize communities. Rare disease increases this responsibility because combinations of phenotypic and genomic features can be identifying. Communication matters. Aggregate public evidence should not be narrated as though the secondary analyst recruited the cohort, generated the assay, or established a clinical finding. Source investigators and participants are part of the evidence history, and attribution should reflect that. Quality control must follow the question A generic quality-control pipeline is not enough. The relevant checks depend on the claim: coverage and transcript consequence for rare variants; donor composition and pseudoreplication for single-cell contrasts; ontology release and term influence for phenotype similarity; tissue handling and postmortem variables for state programs. Negative controls should also be source aware. A control signature derived from fresh blood may not diagnose dissociation effects in brain tissue. A frequency threshold appropriate for one ancestry or inheritance model may be misleading in another. Primary-evidence discipline means matching the control to the generating process. Release what another group needs to disagree A reusable release includes source manifests, environment locks, transformation records, analysis code, sensitivity summaries, and interpretive limits. When data access prevents redistribution, the release should still provide enough identifiers and checks for an authorized group to reconstruct the work. The purpose is not ceremonial openness. It is adversarial usability: another group should be able to rerun the analysis, substitute a source, challenge a mapping, and understand why the conclusion changes. Editorial standard Positions we apply in review Public availability does not remove study design or consent context. Donor and cohort structure must remain visible after integration. Harmonization should preserve disagreement and mapping confidence. Minimum-necessary use and non-reidentification are baseline obligations. A public-data release should enable reconstruction and informed disagreement. REELD conclusion Our conclusion Our editorial conclusion is that public-data science should be held to the same inferential and ethical standards as newly generated evidence. REELD will treat provenance as part of the result, use the biological replicate as the unit of inference, preserve source boundaries, and narrow claims when metadata cannot support more. Continue exploring Read more editorials REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # From target nomination to therapeutic hypothesis | REELD URL: https://reeld.org/editorials/from-target-to-hypothesis Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All editorials Translational Editorial May 28, 2026 18 min read From target nomination to therapeutic hypothesis A ranked gene is the start of the argument. Cell context, direction of effect, and intervention logic complete it. Editorial scope This is a REELD institutional perspective intended to support rigorous scientific review. It has not undergone external journal peer review. Review questions What exact molecular quantity or activity should change? In which cell state and developmental window? Does the intervention modality match the allelic mechanism? What functional readout connects target engagement to phenotype? Which result would invalidate or reverse the proposed direction? Our position A ranked gene is not a therapeutic hypothesis. It identifies where to look. Translation begins only when the argument specifies the causal variant or perturbation class, the molecular direction, the relevant cell state, the phenotypic consequence, the intervention point, and the measurement that would disconfirm the chain. This standard is especially important in rare disease. The same gene can support gain-of-function, dominant-negative, haploinsufficient, recessive, tissue-specific, or developmental mechanisms. A target-level recommendation that ignores allelic direction can propose an intervention opposite to the required correction. Build the chain without collapsing its links The evidence chain runs from variant consequence to molecular effect, pathway or program, cell type and state, tissue physiology, phenotype, and intervention. Each link should carry source evidence, confidence, competing explanations, and transport boundary. A composite score can prioritize the chain but should not erase where it is weak. We prefer an evidence graph because disagreement is scientifically informative. Strong gene–disease validity can coexist with uncertain variant consequence. Strong cell-state localization can coexist with unknown molecular direction. Strong pathway evidence can coexist with an intervention that cannot reach the relevant developmental window. Direction of effect is not optional A therapeutic hypothesis must specify whether activity, abundance, localization, timing, interaction, or state should increase, decrease, normalize, or be redirected. “Modulate the pathway” is not sufficient. The desired direction may differ by variant class, cell state, or disease stage. Human genetic evidence is powerful but must be interpreted mechanistically. Population loss-of-function tolerance, disease-associated truncating variants, activating missense variants, and protective alleles do not imply the same intervention. When direction is unresolved, the responsible conclusion is that target nomination precedes therapeutic logic. Cell context determines whether a direction is meaningful A molecular correction can be beneficial in one cell type and harmful in another. Developmental disorders can reflect a transient window that is no longer accessible when symptoms appear. Immune programs can be protective early and damaging when sustained. Cell-state and temporal specificity are therefore not decorative annotations around a target; they are part of the intervention definition. Single-cell atlases can identify candidate contexts, but expression is not target engagement and co-expression is not causality. The atlas narrows the experiment. Functional models must then test whether the predicted state is necessary, sufficient, and reachable. Intervention modality must match the mechanism Gene replacement, transcript reduction, splice correction, protein stabilization, enzymatic inhibition, agonism, degradation, pathway bypass, and state modulation act at different links in the chain. The correct modality depends on allelic mechanism, dosage window, tissue access, reversibility, and safety margin. A plausible target can fail as a therapeutic hypothesis when the required correction is too narrow, the relevant cells are inaccessible, developmental timing has passed, or the pathway lacks a tolerable window. These are not downstream commercial questions. They are biological constraints that belong in early interpretation. Define the experiment that can say no A useful experiment includes a mechanism-matched perturbation, appropriate cellular context, disease-relevant and orthogonal readouts, negative controls, and rescue. It also names the result that would weaken the hypothesis: wrong direction, absent state specificity, failure of rescue, non-specific toxicity, or dissociation between molecular correction and phenotype. Experiments designed only to confirm target engagement cannot validate the full mechanism chain. A pathway marker can move while the disease-relevant cellular function remains unchanged. Translation requires both molecular and phenotypic consequence. Use graded language We distinguish target association, mechanism hypothesis, intervention hypothesis, preclinical support, and clinical evidence. These are not rhetorical variants. They represent different completed links and different uncertainty. A ranked gene should not inherit the language of a validated therapy. Graded language protects scientific decision making. It allows teams to invest in the experiment most likely to reduce uncertainty without pretending that every computational priority is equally mature. Editorial standard Positions we apply in review A therapeutic hypothesis requires allelic and molecular direction. Cell type, cell state, and developmental timing are part of the target definition. The intervention modality must match the proposed mechanism. Target engagement and disease-relevant functional rescue are separate requirements. Every translational claim should name a result that would weaken it. REELD conclusion Our conclusion Our editorial conclusion is that target nomination is the beginning, not the endpoint, of translation. REELD will call a hypothesis therapeutic only when it links variant mechanism, direction, cell context, intervention modality, functional readout, and falsification. Anything less is a research priority and should be labeled honestly. Continue exploring Read more editorials REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Publish the negative space around a result | REELD URL: https://reeld.org/editorials/publish-the-negative-space Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All editorials Open Science Editorial April 9, 2026 15 min read Publish the negative space around a result Failed replications, null tissues, and unstable rankings define where a mechanism does not travel. Editorial scope This is a REELD institutional perspective intended to support rigorous scientific review. It has not undergone external journal peer review. Review questions Was the null condition measured with adequate sensitivity? Are failed transfers biologically comparable? Which analytical choices change the top-level conclusion? Did a negative control refute the result or refine it? Can another group see the tested boundary without reconstructing every intermediate file? Our position Positive enrichment is easier to narrate than the conditions under which it disappears. Yet those conditions often determine whether another group can use the result. A mechanism without its null tissues, failed transfers, unstable parameters, and incompatible phenotypes is a map without boundaries. REELD treats negative space as part of the primary result. We mean more than publishing non-significant p-values. We mean documenting where the signal was measured adequately and not observed, where the analysis lacked power, where a control explained the effect, and where an independent reference changed the interpretation. Not all nulls mean the same thing A null result can represent true absence, inadequate power, low feature detection, incompatible measurement, wrong timing, incorrect cell context, poor annotation, or a model that absorbed the signal. These explanations have different scientific implications. “Not significant” is not a sufficient description. An interpretable null requires an estimand, expected direction, measurement sensitivity, uncertainty interval, quality controls, and comparison with a positive or technically informative reference. If the assay could not have detected a biologically relevant effect, the result defines a measurement limit rather than a mechanism boundary. Failed transfer is a localization tool When a signal appears in one atlas and not another, the first question should not be which atlas is correct. Differences in tissue, developmental stage, donor composition, platform, disease state, and cell taxonomy may localize the effect. The failed transfer can identify the condition required for the mechanism. This requires source-aware comparison. If two references do not support the same biological contrast, non-replication should not be framed as contradiction. The transfer test must distinguish biological restriction from incompatible design. Parameter instability is evidence about dependence A ranking that changes under a reasonable ontology depth, background universe, normalization, or label resolution is telling us what the result depends on. The correct response is to identify the responsible candidate, term, gene set, or source—not to average across settings until the instability disappears. We favor stability maps that show which conclusions persist, which reverse, and which are unresolved. This preserves local structure. A global similarity statistic can look reassuring while the top experimental priority changes completely. Controls can convert a positive into a narrower positive A negative control that explains part of a signal does not necessarily invalidate the study. It can refine the claim. A disease module that overlaps generic interferon response may retain a smaller disease-specific residual. A cell-state enrichment explained by expression level may still contain a context-specific neighborhood after matching. Publishing only the raw positive or only the post-control residual hides the scientific transition. Both belong in the record, together with the reason the claim changed. Negative space changes resource allocation Experimental follow-up is expensive. If a computational mechanism fails in adult tissue, requires a developmental window, or disappears under donor-aware inference, those facts should influence whether and how a model is built. Hidden negative space causes other groups to repeat avoidable failures or select the wrong context. The value is cumulative. Well-characterized failures can support meta-analysis, benchmark development, power calculations, and better control libraries. Unreported failure preserves a cleaner story for one paper and a dirtier evidence base for the field. A release format for boundaries We recommend a boundary table for every mechanism claim: tested source or condition, expected signal, observed direction and uncertainty, measurement adequacy, interpretation, and consequence for scope. Sensitivity outputs should be machine readable where possible so another group can compare boundaries across studies. Licensing and governance can limit what is released. They do not prevent a clear description of which tests were performed and what they imply. Editorial standard Positions we apply in review A null result requires an interpretable measurement context. Failed transfer can localize a mechanism rather than simply refute it. Parameter instability should be traced, not averaged away. Controls often narrow a claim and that change should be visible. Boundary tables belong beside primary findings. REELD conclusion Our conclusion Our editorial conclusion is that the negative space around a result is part of its meaning. REELD will publish where a signal fails, why the test was informative or limited, and how the failure changes the claim. A bounded mechanism is more transferable than an apparently universal mechanism whose boundaries were never tested. Continue exploring Read more editorials REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Choosing a cell atlas without choosing your conclusion | REELD URL: https://reeld.org/insights/choosing-a-cell-atlas Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All insights Methods August 11, 2026 REELD lab note Choosing a cell atlas without choosing your conclusion A practical checklist for tissue coverage, donor structure, annotation depth, and replication. In practice Use this note as an analytical checklist, then adapt it to the source data and research question. Start with the biological question An atlas should be selected because it samples the tissue, developmental period, and cell resolution required by the mechanism. Dataset size is not a substitute for biological fit. Audit the study design Record donor count, disease status, tissue handling, sequencing protocol, and annotation method. These details define which contrasts the atlas can support. Plan replication before analysis Select at least one independent reference before examining the main result. This reduces the temptation to choose a validation dataset that flatters the first finding. Continue exploring More lab notes REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # What gene constraint can and cannot say | REELD URL: https://reeld.org/insights/what-gene-constraint-can-say Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All insights Rare variants July 31, 2026 REELD lab note What gene constraint can and cannot say Constraint is strong population evidence, but it is not a disease mechanism by itself. In practice Use this note as an analytical checklist, then adapt it to the source data and research question. A population signal Constraint measures depletion of classes of variation relative to an expectation model. It can support dosage sensitivity or selection, but it does not identify a specific disease mechanism. Context still matters A constrained gene may be important across many tissues. Tissue and cell-state evidence helps determine whether that general importance is relevant to the phenotype under study. Use it as one layer Report constraint alongside inheritance, variant consequence, phenotype fit, and functional context. Avoid allowing one summary metric to dominate the full evidence graph. Continue exploring More lab notes REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # When an ontology release changes your result | REELD URL: https://reeld.org/insights/ontology-release-drift Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All insights Phenotypes July 3, 2026 REELD lab note When an ontology release changes your result Version drift can alter term relationships and ranking behavior. Here is how we audit it. In practice Use this note as an analytical checklist, then adapt it to the source data and research question. Relationships change Ontology releases add terms, revise definitions, and alter parent-child relationships. A semantic similarity result can therefore change without any new patient observation. Lock and compare Record the release used in every analysis and periodically compare results with the current release. A changed ranking should be traced to the responsible term or relationship. Publish the delta When an update matters, report both the old and new representation. This makes the effect of knowledge-base evolution visible. Continue exploring More lab notes REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Building a negative-control signature library | REELD URL: https://reeld.org/insights/negative-control-library Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All insights Single cell June 18, 2026 REELD lab note Building a negative-control signature library Common stress, cell-cycle, interferon, and dissociation programs belong in every enrichment workflow. In practice Use this note as an analytical checklist, then adapt it to the source data and research question. Why controls belong early Single-cell enrichment is vulnerable to programs that recur across many tissues and conditions. Stress, cell cycle, interferon response, and dissociation effects can produce convincing but nonspecific signal. Build a reference set Maintain versioned signatures for common technical and biological programs. Test the target gene set against these controls using the same scoring and resampling procedure. Interpret overlap Overlap does not automatically invalidate a result. It changes the claim. A disease program may include a common response, but the specific component must be identified. Continue exploring More lab notes REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Why we prefer an evidence graph to one score | REELD URL: https://reeld.org/insights/evidence-graph-not-score Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All insights Interpretation May 16, 2026 REELD lab note Why we prefer an evidence graph to one score One number hides disagreement. A graph preserves which evidence supports each step. In practice Use this note as an analytical checklist, then adapt it to the source data and research question. Scores compress disagreement A composite score can rank candidates while hiding whether evidence layers agree. Two genes with the same score may have entirely different support profiles. Graphs preserve the argument An evidence graph keeps each source, inference, and uncertainty connected but inspectable. Reviewers can see which link supports a proposed mechanism. Ranking remains possible A graph does not prevent prioritization. It makes the basis of a ranking explicit and allows alternate weighting without rebuilding the evidence record. Continue exploring More lab notes REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # A minimum provenance record for public data | REELD URL: https://reeld.org/insights/public-data-provenance Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate All insights Open science April 24, 2026 REELD lab note A minimum provenance record for public data Source accessions, versions, transformations, exclusions, and checks every analysis should retain. In practice Use this note as an analytical checklist, then adapt it to the source data and research question. Minimum source record Retain source accession, download date, release, license or access conditions, and any source-provided sample exclusions. Transformation record Record filtering, normalization, mapping references, label changes, and software environment. A notebook alone is rarely a sufficient provenance system. Output identity Tie every derived result to input manifests and code state. Checksums or equivalent immutable identifiers make silent file changes detectable. Continue exploring More lab notes REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Variant to phenotype | REELD URL: https://reeld.org/research/variant-to-phenotype Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Research programs Variant to phenotype A rare variant becomes informative when consequence, gene biology, inheritance, and phenotype form a coherent evidence chain. Variant consequence Compare transcript-aware annotations and retain disagreement between sources. Gene-level priors Use constraint and disease association as calibrated evidence, not automatic answers. Phenotype fit Model exact terms, informative ancestors, and uncertainty in observation. Questions we force the analysis to answer Every output is evaluated against explicit interpretive tests. Which transcript and consequence definitions change the interpretation? Does the candidate remain plausible when broad phenotype terms are removed? What evidence would distinguish causal biology from coincidental overlap? Related studies Benchmark study Cardiovascular genetics Constraint-aware prioritization in pediatric cardiomyopathy Integrating population constraint without allowing it to overwhelm tissue and phenotype evidence. Read study Replication study Functional genomics Cross-atlas replication of dosage-sensitive programs A study of which haploinsufficiency signals persist across public single-cell references. Read study REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Single-cell context | REELD URL: https://reeld.org/research/single-cell-context Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Research programs Single-cell context Disease mechanisms become actionable when they are placed in the right cell type, state, and developmental window. Cell identity Harmonize labels across references without pretending that taxonomies are identical. Cell state Separate constitutive expression from disease-relevant activation and maturation programs. Cross-atlas transfer Require signals to persist beyond one cohort, protocol, or annotation scheme. Questions we force the analysis to answer Every output is evaluated against explicit interpretive tests. Is the signal specific to a cell state or common across the tissue? Does it survive alternate preprocessing and label resolution? Which perturbation context can directly test the proposed mechanism? Related studies Methods study Cardiovascular genetics Cell-state restriction clarifies MYH7 variant signals A reproducible workflow for asking when cardiomyopathy-associated variation becomes biologically legible. Read study Atlas study Neurodevelopment Convergent phenotype modules across neurodevelopmental genes Testing whether distinct rare-disease genes converge on shared developmental programs. Read study REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Phenotype ontology | REELD URL: https://reeld.org/research/phenotype-ontology Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Research programs Phenotype ontology Phenotypes are not a static list. Their structure, specificity, and absence shape every downstream result. Representation Record exact terms, modifiers, exclusions, and ontology release. Semantic structure Test ancestor propagation and information-content choices. Ranking stability Measure which terms dominate candidate prioritization and why. Questions we force the analysis to answer Every output is evaluated against explicit interpretive tests. Which terms carry genuine discriminating information? How does ontology version change gene or disease similarity? What happens when uncertain or absent phenotypes are modeled explicitly? Related studies Benchmark study Cardiovascular genetics Constraint-aware prioritization in pediatric cardiomyopathy Integrating population constraint without allowing it to overwhelm tissue and phenotype evidence. Read study Replication study Functional genomics Cross-atlas replication of dosage-sensitive programs A study of which haploinsufficiency signals persist across public single-cell references. Read study REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy --- # Therapeutic hypotheses | REELD URL: https://reeld.org/research/therapeutic-hypotheses Skip to content REELD will exhibit at ASHG 2026 — October 20–24, Montréal. Meet our team at the meeting REELD Research Studies Case studies Editorials About Our Story Collaborate Research programs Therapeutic hypotheses A useful hypothesis specifies cellular context, direction of correction, expected readout, and a reason it could be wrong. Mechanism chain Connect variant effect to molecular program, cell state, and phenotype. Intervention logic Define whether a program should be restored, reduced, redirected, or timed differently. Validation plan Name the assay, context, readout, and disconfirming observation. Questions we force the analysis to answer Every output is evaluated against explicit interpretive tests. Is direction of effect supported at the variant and pathway levels? Can the intervention reach the relevant cell state and developmental window? Which result would force the hypothesis to be revised? Related studies Benchmark study Cardiovascular genetics Constraint-aware prioritization in pediatric cardiomyopathy Integrating population constraint without allowing it to overwhelm tissue and phenotype evidence. Read study Replication study Functional genomics Cross-atlas replication of dosage-sensitive programs A study of which haploinsufficiency signals persist across public single-cell references. Read study REELD Computational rare-disease biology built for reproducibility, challenge, and translation. REELD is a non-profit educational research institute. lab@reeld.org Research Programs Studies Methods Data practice Institute About Our Story News Collaborate Transparency Contact Writing Case studies Editorials Insights Library Research digest Join Monthly. Methods, studies, and critical reviews. © 2026 REELD.ORG · Founded 2016 REELD is a non-profit educational research institute. Research use only. Not clinical guidance. Privacy ---