To determine the deleteriousness of missense variants, we identified orthologs for each gene containing a missense mutation using the Ensembl compara database (v. 47), requiring a one-to-one relationship and minimum 80% amino acid identity to ensure good alignments. The protein and its orthologs were aligned using MAFFT40 version 6.240, ‘linsi’ method, with 1,000 iterations. The alignment was scored at the missense position for conservation using Scorecons41, with scoring method ‘valdar01’, matrix ‘modified PET91’, and matrix transformation ‘karkinlike’. The conservation score for an amino acid affected by a missense change was calculated as the product of the Scorecons score, representing how identical the orthologs were at the given position, and the number of orthologs for the gene, giving an overall conservation score that is higher for amino acids that are highly conserved deeply in evolution. The efficacy of the deleteriousness score is illustrated by a comparison of nonrecurrent and recurrent missense variants (Supplementary Fig. 2 online) Additionally, a score for amino acid change was determined using a Grantham matrix42.
