Analysis plan fixed and deposited before any AlphaGenome Atlas score was retrieved for the variants it covers. It extends a within-site audit of AlphaGenome Atlas predictions at ClinVar pathogenic/likely pathogenic sites from four genes (DNM1, PCCA, PCCB, MMUT) to the seven further genes of a curated 12-gene panel (SCN5A, BAP1, HFE, ASS1, TSC1, BRCA2, RB1; 2,144 positions, 6,432 substitutions). It fixes the design, the analysis and fetch code (by SHA-256), the known-answer controls, six readings (replication of the nonsense-versus-missense contrast, for AVI and for scorers without consequence inputs; a confirmatory consequence-matched test; the missense question under a context-aware mutation null; the intronic within-class signal; canonical splice positions), sensitivity analyses and stopping rules. The deposit date is the evidence that the plan predates the scores. By using this information, you agree to AlphaGenome Output Terms of Use found at http://deepmind.google.com/science/alphagenome/output-terms. Summary counts in section 1 are derived from AlphaGenome Atlas outputs; the plan itself contains no Atlas output.
No takes yet. Share an insight, caveat, or question.
Patrick Iben (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: