AlphaMissense community resource
71 million missense variant predictions covering the human proteome -
missense proteome data
For each UniProt canonical isoform, first generate all possible SNVs and then keep SNVs that change the protein seq
Gene-level AlphaMissense pathogenicity predictions
which is the average pathogenicity of all possible missense variants in a gene
216 million possible single amino acid substitutions across 19,233 human proteins (UniProt canonical isoforms)
Important
Missense proteome data :tada:
Out of 71 million missense variant covering the human proteome
Likely pathogenic: 32% (22.8 million) are and
Likely benign: 57% (40.9 million)
Classification cutoff: 90% precision using the ClinVar dataset
Applications
AlphaMissense findings coupled with donstream functional experiments improve the current understanding of clinically actionable genes and variants
Expedite understanding of the molecular effects of variants on protein function.
Prioritise rare disease relevant variants
Improve the diagnostic yield of rare genetic diseases