Pipeline
The preprocessing step reads the seed mutation catalog, normalizes records into a typed schema, builds facet summaries, ranks variant dossiers by evidence density, and writes static artifacts under public/atlas-data. The public Explore interface loads compact search shards only when the user enters Explore.
Record model
Core records preserve gene, HGVS labels, genomic positions, cancer type mappings, source databases, citations, recurrence, hotspot status, actionability fields, drug and trial hooks, clinical labels, and outbound links.
Freshness
Seed source generated: Feb 4, 2026. Atlas artifacts generated: Apr 19, 2026.
Validation stance
The v1 build performs schema normalization, duplicate slug protection, required-field coverage reporting, facet generation, and route-payload stability tests. Missing data is reported as missing, not inferred.