{ "registry_schema": "nmd-vcell-reliability-advantage-registry/1.1", "registry_schema_url": "https://nmdvcell.com/resource/api/v1.1/reliability_advantage_registry.schema.json", "resource_release": "v1.2.0-measured-dmd-evidence", "interface_build": "EA-20260817-57", "assessed_at": "2026-08-16", "route": "/resource/reliability/", "overall_assessment": "Reliable for evidence identity, provenance, bounded synthesis and research workflow; limited for the frozen same-context aggregate benchmark; not established for cross-context DMD response; prohibited for therapeutic, patient or clinical use.", "scoring_policy": "No composite reliability score is emitted because engineering integrity, evidence synthesis, perturbation prediction and clinical validity are non-interchangeable claims.", "advantage_claims": [ { "advantage_id": "ADV-01", "title": "Evidence boundaries are product behavior", "claim": "Measured, contextual, modeled, missing, failed and prohibited states remain distinct on public objects.", "proof": [ "typed evidence states", "claim ceilings", "explicit abstentions", "empty registries remain visible" ], "current_strength": "PROVEN_PRODUCT_CAPABILITY", "durability": "HIGH_IF_GOVERNANCE_REMAINS_ENFORCED", "weakness": "The advantage disappears if future releases collapse missingness into scores or hide failed runs.", "falsification_test": "Audit every new result route for source, task, execution state, prohibited uses and missing target-context outcome.", "next_investment": "Make the reliability registry a required release gate.", "evidence_routes": [ "/resource/claim_boundary", "/resource/models/", "/resource/registry" ], "evidence_object_ids": [ "RELEASE:v1.2.0-measured-dmd-evidence:EA-20260817-57", "MRUN-RIDGE-SAFE-2.3-G0-REPEATED-FOLD" ] }, { "advantage_id": "ADV-02", "title": "Neuromuscular context is first-class", "claim": "Disease, cell state, perturbation, time and phenotype are represented as the scientific unit rather than collapsed into a gene score.", "proof": [ "four disease routes", "DMD process graph", "cell-context registry", "21 governed DMD decision contracts" ], "current_strength": "PROVEN_RESOURCE_CAPABILITY", "durability": "MEDIUM_UNTIL_PROPRIETARY_PERTURBATION_TRUTH_EXISTS", "weakness": "Current DMD assets are rich context but do not contain matched candidate-conditioned DMD response truth.", "falsification_test": "Verify that every disease claim names the exact source context and inferential unit.", "next_investment": "Acquire a matched DMD myogenic perturbation pilot with independent biological units.", "evidence_routes": [ "/resource/diseases/", "/resource/process/", "/resource/cell-browser/" ], "evidence_object_ids": [ "MDEMANIFEST:v1.2.0-measured-dmd-evidence:EA-20260817-57", "PROJECTSTATE:v1.2.0-measured-dmd-evidence:DMD-MIN-PERTURBOME:1.1" ] }, { "advantage_id": "ADV-03", "title": "Research objects form a closed loop", "claim": "Dataset, ModelRun, Study, Prediction and Outcome share identities and legal handoffs.", "proof": [ "Dataset Cards", "immutable ModelRun releases", "Study Card planner", "prospective lifecycle registry" ], "current_strength": "STRUCTURALLY_IMPLEMENTED_NO_RETURNED_OUTCOMES", "durability": "HIGH_AFTER_FIRST_PROSPECTIVE_OUTCOME_RETURN", "weakness": "The outcome registry is intentionally empty, so the loop is structurally complete but not yet empirically exercised.", "falsification_test": "A future outcome must trace back to its frozen study, prediction, data and decision rule without manual reconstruction.", "next_investment": "Register and return the first prospective null, positive, toxic or failed outcome.", "evidence_routes": [ "/resource/datasets/", "/resource/models/", "/resource/planner", "/resource/registry" ], "evidence_object_ids": [ "LIFECYCLE:v1.2.0-measured-dmd-evidence:EA-20260817-57", "MRUN-RIDGE-SAFE-2.3-G0-REPEATED-FOLD" ] }, { "advantage_id": "ADV-04", "title": "Simple baselines are permanent", "claim": "Advanced models cannot advance by reputation, scale or one aggregate metric.", "proof": [ "same-coverage baseline rule", "frozen G0-G7 tasks", "six-level evaluation ladder", "GEARS and scGPT failures retained" ], "current_strength": "PROVEN_METHOD_GOVERNANCE", "durability": "HIGH_IF_BASELINES_REMAIN_MANDATORY", "weakness": "A baseline firewall protects decisions but does not create missing disease outcome.", "falsification_test": "Re-run each advanced model against identical data, split, genes, seeds and metric code.", "next_investment": "Add perturbation-specific and population baselines as new output types become executable.", "evidence_routes": [ "/resource/technology-methods/", "/resource/benchmarks/", "/resource/models/" ], "evidence_object_ids": [ "MRUN-RIDGE-SAFE-2.3-G0-REPEATED-FOLD", "MRUN-GEARS-0.1.2-FIVE-SEED-20260713", "MRUN-SCGPT-0.2.5-FIVE-SEED-20260714" ] }, { "advantage_id": "ADV-05", "title": "Negative results remain reusable assets", "claim": "Failed, no-go, stopped, null and abstained states are released with the same identity discipline as positive results.", "proof": [ "six ModelRun release pages", "reproduction receipts", "null cross-disease results", "prohibited-use fields" ], "current_strength": "PROVEN_PRODUCT_CAPABILITY", "durability": "HIGH_IF_SELECTION_BIAS_IS_RESISTED", "weakness": "Negative computational results do not replace prospective biological validation.", "falsification_test": "Compare public registries against the complete execution ledger and require every terminal run to be represented.", "next_investment": "Attach resource-cost and failure-taxonomy fields to future ModelRuns.", "evidence_routes": [ "/resource/models/", "/resource/model-run/mrun-gears-0-1-2-five-seed-20260713/", "/resource/diseases/compare/" ], "evidence_object_ids": [ "MRUN-GEARS-0.1.2-FIVE-SEED-20260713", "MRUN-SCGPT-0.2.5-FIVE-SEED-20260714", "MRUN-TXPERT-CONFIG-GAT-SEED-20260712", "MRUN-MORPH-DEPMAP25Q3-VALIDATION-20260722" ] } ], "reliability_domains": [ { "domain_id": "REL-01", "title": "Software & release integrity", "state": "HIGH_ENGINEERING_CONFIDENCE", "short_reason": "Build, route, schema and cross-surface gates pass locally.", "evidence": [ "full vinext production build PASS", "51 rendered contract tests", "246 multidisease checks", "24 cross-surface checks", "checksum-addressed release objects" ], "evidence_object_ids": [ "RELEASE:v1.2.0-measured-dmd-evidence:EA-20260817-57" ], "limitation": "Anonymous command-line production QA is blocked by the existing edge HTTP 403 policy.", "permitted_use": "Rebuild, audit and navigate the released research resource.", "prohibited_inference": "Engineering integrity does not validate a biological prediction.", "next_gate": "Add authenticated post-deploy route checks to the immutable production receipt." }, { "domain_id": "REL-02", "title": "Identity, provenance & data contracts", "state": "HIGH_RESOURCE_CONFIDENCE", "short_reason": "Identifiers, sources, contexts, nulls and transformations are explicit.", "evidence": [ "17,921 released gene records", "HGNC alias map", "Dataset Cards", "typed API", "checksums and release identities" ], "evidence_object_ids": [ "NMDVCELL-HEPG2-CRISPRI", "GSE293514", "GSE272233", "GSE52529", "GSE133344" ], "limitation": "A valid data object can still be unsuitable for a particular prediction task.", "permitted_use": "Identity resolution, evidence lookup and reproducible data intake.", "prohibited_inference": "Presence in the resource is not disease relevance or efficacy.", "next_gate": "Require ontology and inferential-unit validation on every newly imported dataset." }, { "domain_id": "REL-03", "title": "Disease evidence synthesis", "state": "BOUNDED_RESEARCH_CONFIDENCE", "short_reason": "Source-linked context is strong enough for hypothesis and study design.", "evidence": [ "DMD measured-evidence release", "four disease modules", "cell-context map", "process and evidence graphs" ], "evidence_object_ids": [ "MDEMANIFEST:v1.2.0-measured-dmd-evidence:EA-20260817-57", "PROJECTSTATE:v1.2.0-measured-dmd-evidence:DMD-MIN-PERTURBOME:1.1" ], "limitation": "Observational and correction context does not establish candidate-conditioned causal response.", "permitted_use": "Mechanism navigation, gap analysis and prospective study design.", "prohibited_inference": "Do not treat integrated context as target efficacy or causal proof.", "next_gate": "Independent replication plus matched perturbation and functional endpoints." }, { "domain_id": "REL-04", "title": "Same-context perturbation benchmark", "state": "LIMITED_TASK_CONFIDENCE", "short_reason": "Ridge slightly improves mean RMSE, but response direction is unsupported.", "evidence": [ "MRUN-RIDGE-SAFE-2.3-G0-REPEATED-FOLD", "RMSE 0.1175 vs train mean 0.1180", "strict outcome gate 0/16" ], "evidence_object_ids": [ "MRUN-RIDGE-SAFE-2.3-G0-REPEATED-FOLD" ], "limitation": "The output is a 2,000-feature aggregate HepG2 mean vector, not a cell population or DMD response.", "permitted_use": "Permanent same-assay comparator and leakage diagnostic.", "prohibited_inference": "No directional transfer, DMD prediction or functional effect.", "next_gate": "Beat matched means on perturbation-specific delta and distribution metrics across prospective holdouts." }, { "domain_id": "REL-05", "title": "Cross-context & DMD response prediction", "state": "NOT_ESTABLISHED", "short_reason": "No model has passed a matched DMD candidate-response task.", "evidence": [ "GEARS failed five frozen splits", "scGPT lost to same-coverage baselines", "TxPert calibration pending", "direct DMD candidate outcome 0/21" ], "evidence_object_ids": [ "MRUN-GEARS-0.1.2-FIVE-SEED-20260713", "MRUN-SCGPT-0.2.5-FIVE-SEED-20260714", "MRUN-TXPERT-CONFIG-GAT-SEED-20260712", "LIFECYCLE:v1.2.0-measured-dmd-evidence:EA-20260817-57" ], "limitation": "The required donor- and cell-state-matched perturbation outcome matrix does not exist in the release.", "permitted_use": "Method research, task design and governed abstention.", "prohibited_inference": "No candidate ranking, efficacy prediction or virtual DMD response population.", "next_gate": "Acquire and freeze matched DMD perturbation outcomes before model selection." }, { "domain_id": "REL-06", "title": "Therapeutic, patient & clinical use", "state": "PROHIBITED", "short_reason": "No clinical calibration, prospective outcome set or authorization exists.", "evidence": [ "zero calibrated DMD models", "zero prospective predictions", "zero measured outcomes", "DOI and institutional licences pending" ], "evidence_object_ids": [ "LIFECYCLE:v1.2.0-measured-dmd-evidence:EA-20260817-57", "RELEASE:v1.2.0-measured-dmd-evidence:EA-20260817-57" ], "limitation": "The platform is a research resource, not a medical device or treatment recommender.", "permitted_use": "None beyond research hypothesis and study-design support.", "prohibited_inference": "No diagnosis, prognosis, treatment selection, dosing or patient-level digital twin.", "next_gate": "A separate clinically governed development and validation program; current research gates cannot unlock it." } ], "moat_plan": [ { "stage_id": "MOAT-00", "title": "Trust infrastructure", "state": "IMPLEMENTED", "deliverable": "Typed evidence, model and reliability contracts with negative-result preservation.", "acceptance_gate": "Every claim has provenance, execution state, failure state, ceiling and next gate.", "no_shortcut": "Do not replace explicit states with a synthetic trust score.", "advantage_unlocked": "Auditable research decision support." }, { "stage_id": "MOAT-01", "title": "Matched DMD perturbation pilot", "state": "BLOCKED_EXTERNAL_EXPERIMENT", "deliverable": "Candidate-conditioned DMD myogenic perturbation outcomes with independent biological units, molecular and functional readouts.", "acceptance_gate": "Frozen estimand, controls, donor/culture units, QC, toxicity and complete outcome return.", "no_shortcut": "Do not substitute healthy myoblast, HepG2 or unperturbed DMD context.", "advantage_unlocked": "First proprietary disease-task truth and model eligibility." }, { "stage_id": "MOAT-02", "title": "Donor-resolved benchmark", "state": "PENDING_MOAT_01", "deliverable": "Leakage-safe donor, state and perturbation holdouts with permanent simple baselines.", "acceptance_gate": "Perturbation-specific, population, calibration and functional layers pass prospectively.", "no_shortcut": "Cells cannot be treated as independent donors and mean RMSE cannot be the only metric.", "advantage_unlocked": "Defensible NMD generalization evidence." }, { "stage_id": "MOAT-03", "title": "Prospective learning loop", "state": "PENDING_MOAT_02", "deliverable": "Registered predictions whose positive, null, toxic and failed outcomes return to the same evidence graph.", "acceptance_gate": "Immutable preregistration, blinded evaluation, calibration and independent replication.", "no_shortcut": "Retrospective fit cannot be relabeled as prospective validation.", "advantage_unlocked": "Compounding proprietary evidence and a genuinely exercised closed loop." } ], "release_receipt": { "source_candidate": { "state": "UNVERIFIED_AFTER_SOURCE_BUILD", "required_gates": "Full build, scientific validators, rendered contracts and lint", "deployment_receipt": "EXTERNAL_POST_BUILD_RECEIPT_REQUIRED" }, "latest_bundled_historical_deployment": { "sites_version": 167, "deployment_status": "SUCCEEDED", "content_commit": "91e7e8ff973d47db9f010d89879276000b5e2ffa", "production_url": "https://nmdvcell.com/resource/", "provider_url": "https://nmd-vcell.sema3c.chatgpt.site/resource/", "deployed_at": "2026-09-04T12:36:05.613749+00:00", "anonymous_route_qa": "BLOCKED_HTTP_403", "receipt_scope": "COMPLETED_PRIOR_DEPLOYMENT_NOT_CURRENT_SOURCE_CANDIDATE" } }, "valid_for_release": true, "update_rule": "A reliability state changes only when its named next gate produces a versioned object; a new page, model name or external citation cannot upgrade it.", "claim_boundary": "This registry audits current product and evidence reliability. It is not a certification, medical-device assessment, therapeutic recommendation or substitute for prospective biological and clinical validation." }