ABSTRACT
This chapter concentrates on the computational pathogenicity prediction for variants within transcripts of protein-coding genes. It presents some of the main features that many of the programs use for their prediction. Pathogenicity prediction exploits certain attributes of the affected genes, transcripts, transcript positions, and the sequence context of the nucleotide and amino-acid alterations associated with variants in order to assess whether a given variant is disease-causing or not. For most programs, the predictions are best interpreted as attempting to predict not disease causality but deleteriousness of the variant for protein function: a variant may completely abolish the function of a gene that is simply not associated with any disease. Variants that affect protein-coding genes are mainly assessed with reference to their predicted effects on the amino acid sequence, structure, or function of the protein or in some cases with respect to potential changes in the amount of protein produced.
