Infrastructure · Biology · Medicine

AlphaMissense releases proteome-wide missense variant predictions

Proteome-wide missense variant pathogenicity predictions for possible human single-amino-acid substitutions.

Summary

Google DeepMind describes AlphaMissense, an AlphaFold-adapted model fine-tuned on human and primate variant frequency data to score missense variant pathogenicity. The release includes predictions for possible human single amino acid substitutions and classifies 89% of missense variants as likely benign or pathogenic.

AI role

Adapted AlphaFold-style modeling and variant-frequency data to prioritize likely benign or pathogenic missense variants.

Narrative role

This is a useful infrastructure event because it shows AI-generated catalogs becoming research triage layers for genomics and rare-disease work.

Caveat

Variant predictions are not clinical diagnoses and require independent experimental, population, and clinical evidence.