The CodeXome framework does not predict which amino-acid substitutions proteins tolerate. It records which substitutions have already been tested across primate lineages — substitutions that have persisted through millions of generations of purifying selection on biochemically similar proteins.
Every validation study on this page evaluates a specific implication of this framework against an independent clinical or functional reference: ClinGen expert classifications, ENIGMA/BRCA Exchange clinical assertions, ClinVar VUS calls, CFTR therapeutic-response data.
2,689 expert-classified SNVs from 52 ClinGen-curated genes were cross-referenced against the CodeXome primate exome database. Variants classified pathogenic or likely pathogenic by ClinGen expert panels were absent from primate alignments (n = 0 shared). Variants classified benign or likely benign showed recurrence across primate lineages.
Replication of the ClinGen cross-reference protocol on 10,699 BRCA1/2 variants with ENIGMA consortium classifications. Directionality preserved: pathogenic and likely pathogenic variants absent from primate alignments; benign and likely benign variants present. Finding held across a second independent expert-classified dataset.
Missense variants with ClinVar classifications of "uncertain significance" or "conflicting interpretations" were cross-referenced against primate recurrence data across ~14,000 genes. Mean reclassifiable-as-likely-benign fraction: 20%. Per-gene range: 10-40%, observed in 93% of genes tested.
DTAS scores computed for 260 CFTR mutations with established therapeutic response data. Predicted functional-impact class compared against clinical response classification. Accuracy: 99.3%. Benchmark gene selected for density of functional and clinical ground truth.
Running a rare disease reanalysis? Building a variant interpretation pipeline? Pilot partnerships and research trials are open. Reach out to discuss how evolutionary evidence can strengthen your study design.