Every entry starts with a biological or audit question.
Tool Catalog · task-first entry points
Start from the scientific job, not the page hierarchy.
Each tool declares who it is for, the accepted input, returned object, suitability rule, failure states and next handoff. “Live” describes the web workflow—not an unreported model execution.
Inputs and outputs connect to stable platform objects.
Missing data and context mismatch return an explicit next action.
Every result points toward evidence, experiment or outcome.
Seven bounded workflows
Use the platform as a connected research instrument.
No catalog entry inherits external performance or hides an empty registry behind a polished interface.
Gene Resolver
Which canonical gene does this symbol, alias, previous symbol or identifier resolve to?
Use whenSuitable for identifier normalization and navigation; not a biological inference engine.
Failure states & responsible use
IDENTIFIER_NOT_FOUNDNo released mapping key matches the input. Check source species, namespace and spelling.AMBIGUOUS_ALIASThe alias cannot be assigned safely to one canonical symbol. Use Ensembl or NCBI Gene ID.
Do not interpret successful identity resolution as evidence of disease relevance.
Evidence Comparator
How do several genes differ across evidence dimensions without forcing a winner?
Use whenSuitable for side-by-side evidence inspection; not suitable for therapeutic ranking.
Failure states & responsible use
GENE_NOT_ASSESSEDA dimension has no released evidence for one or more genes. Open the gene page and inspect the missing-edge experiment.COMPARISON_NOT_FROZENThe requested set has no immutable comparison object. Create a shareable comparison from the compare route.
Do not convert a comparison into a clinical or therapeutic recommendation.
Process Inspector
Which observed, contextual or hypothesized edges connect a candidate to a DMD process?
Use whenSuitable for mechanism navigation and experiment design; not a causal simulator.
Failure states & responsible use
EDGE_NOT_VALIDATED_IN_DMDThe edge is contextual or hypothesized rather than a validated DMD transition. Use the named missing-edge experiment.
Do not describe a pathway edge as a predicted cell-state transition unless a registered model and matched truth support it.
Model Auditor
What exact task ran, against which baselines, and why did the gate pass, fail or stop?
Use whenSuitable for audit and model selection within the frozen task; external reputation is not accepted as local performance.
Failure states & responsible use
PARTIAL_REPRODUCTION_RECEIPTCode, container, data digest or seed receipt is missing. Do not claim clean-room reproducibility; inspect the explicit missing fields.TASK_CONTEXT_MISMATCHThe run context does not qualify the requested disease prediction. Return to the Dataset Card and task contract.
Completed does not mean calibrated for DMD, therapeutic use or clinical use.
Method Navigator
Which model family, permanent baselines and evaluation layers fit this frozen biological task?
Use whenSuitable for method selection and review; it does not execute a model or transfer literature performance into NMD-VCell.
Failure states & responsible use
TASK_NOT_FROZENThe biological unit, output or generalization axis is undefined. Write the task and split contract before choosing an architecture.TRUTH_CONTEXT_MISSINGNo matched outcome exists for the requested disease and cell state. Return an abstention and open the Experiment Planner.BASELINE_NOT_REGISTEREDThe advanced model has no task-compatible simple comparator. Register zero-change, mean, linear or nearest-neighbor controls first.
Source-reported capability is a method reference, not a local benchmark result or DMD validation.
Experiment Planner
What prospective study would resolve the current decision-blocking evidence gap?
Use whenSuitable for drafting a falsifiable study; the output is not a registered protocol or prediction.
Failure states & responsible use
UNKNOWN_GENEThe candidate cannot be resolved to the released identifier map. Use Gene Resolver first.INCOMPLETE_STUDY_CONTRACTContext, comparator, endpoint or decision rule is missing. Complete all required Study Card fields before review.
A draft Study Card must not be represented as an approved or executed experiment.
Outcome Registry Inspector
Has a draft study become registered, predicted, measured and returned to the evidence graph?
Use whenSuitable for lifecycle audit. The currently empty registries are a truthful state, not missing UI content.
Failure states & responsible use
NO_REGISTERED_OBJECTNo immutable study, prediction or outcome exists for the identifier. Return to the draft planner and registration gate.OUTCOME_NOT_RETURNEDA registered study lacks a measured outcome or evidence-graph return. Complete the governed outcome import.
Do not fabricate outcome rows or infer efficacy from an empty registry.
Operating rule
One tool may hand off to another; none may silently upgrade the claim.
A tool route exposes a governed workflow and its failure states. It does not imply that every requested computation, prediction or outcome is available.