StP: a web server to predict pathogenesis of non-synonymous single nucleotide polymorphisms in membrane proteins

Abstract

Next-generation sequencing has speed up the process of sequencing a complete human genome, while reducing the cost. Consequently its use is increasing in healthcare in order to identify genetic variations and its association with pathology. Single Nucleotide Variants are the most common form of DNA variation in human population and consists in a change in a single nucleotide. One of the current challenges is to be able to predict if a single nucleotide polymorphism can be associated to a single-genetic diseases. Thus, in order to identify the variant among hundreds of variants in a genome that is the cause of a single genetic disease, in silico mutation prediction tools have been developed these lasts years. These tools are based in mathematical, rule-based, and statistical learning methods relying on evolutionary, sequence, or structural information methods to characterize if a residue substitution in proteins is affecting its structure and function. These tools are mainly developed for globular proteins. In order to develop a mutation predictor server specific for membrane proteins, we have developed Single Nucleotide Polymorphisms Transmembrane Predictor (StP). This web server aims to predict if a missense mutation in a transmembrane region of a membrane protein is likely to affect the structure and/or function of a protein and consequently being damaging. The predictive algorithm is based on the entropy of the mutated position, the frequency of the non-mutated amino acid and the frequency of the mutated amino acid computed from Pfam multiple sequence alignments and also on the score associated to the amino acid change. Comparison to existing mutation server shows that StP improves the specificity, although loosing sensitivity in the prediction if a SNP is damaging or not in a membrane protein.