The major principle of the protein-sequence-based methods to predict deleteriousness in the coding sequence is based on comparative genomics and functional genomics. Comparative sequencing analysis assumes that amino acid residues that are critical for protein function should be conserved among species and homologous proteins; therefore, mutations in highly conserved sites are more likely cisplatin synthesis to result in more deleterious effect. Other modalities to predict disease-causing variants include protein biochemistry, such as amino acid charge, the presence of a binding site, and structure information of protein. SNVs that are predicted to alter protein feature (such as polarity and hydropathy) and structure (binding ability and alteration of secondary/tertiary structure) have a higher probability of being deleterious.

Although the majority of research has focused on protein-altering variants, noncoding variants constitute a large portion of human genetic variation. Results obtained from GWAS indicate that ~88% of trait-associated weak effect variants are found in noncoding regions, demonstrating the importance of functional annotation of both coding and noncoding variants [41]. Computational tools for protein-sequence-based prediction of deleteriousness fall into two categories: constraint-based predictors such as MAPP and SIFT, and trained classifiers such as MutationTaster and polyPhen. In addition to protein-sequence-based methods, another way to prioritize disease-casual SNVs is through nucleotide-sequence-based prediction in noncoding and coding DNA.

This process also utilizes comparative genomics to predict deleteriousness, and is used by programs such as phastCons, GERP, and Gumby. In one detailed review of disease-causing variant identification, the authors introduced the concepts and tools that allow genetic annotation of both coding and noncoding variants [39]. They also compared the relative utility of Cilengitide nucleotide- and protein-based approaches using exome data, finding that nucleotide-based constraint scores defined by Genomic Evolutionary Rate Profiling (GERP) and protein-based deleterious impact scores provided by PolyPhen were similar for two Mendelian diseases, suggesting that nucleotide-based prediction can be as powerful as protein-based metrics [39]. Below, we survey tools that are helpful identifying disease-causal variants among numerous candidates.6.1. Sorting Intolerant from Tolerant (SIFT)Sorting Intolerant From Tolerant (SIFT) (http://sift.jcvi.org/) prediction is based on conserved amino acid residues through different species using comparative sequencing analysis through PSI-BLAST [42].