About the resource

Type of resource: Mutation filtering system
Website: Click here to view website
User documentation:
Relevant publication:


With the development of next generation sequencing technologies, the amount of data generated has reached an unprecedent level. The sequencing of human genomes has thus led to the identification of about 3 millions SNPs per individual among which 20-25% are novels (up to 750,000). If we consider data from dbSNP, 86% of variations are localized within introns (645,000) and about 2.8% in exons (21,000). Among these variations 52% correspond to missense variations (10,500) and 37% to synonymous variations (7,770). Considering only pathogenic mutations reported in the HGMD, 56% or reported mutations correspond to missense mutations. Therefore it appears that most variations localized within exons are indeed missense mutations that could either be responsible for a phenotype (pathogenic mutations) or not (non-pathogenic mutation). The ability to distinguish these two classes of mutations is of major interest in medicine as it has a major impact for genetic counseling. We have previously developed the UMD-Predictor algorithm that aims to predict the pathogenicity of any cDNA variation. This algorithm was included in the UMD software. Previous analysis on about 1,000 different substitutions has revealed that this algorithm was the most efficient. We therefore decided to create the UMD-HTS resource to allow the prediction of the best candidate pathogenic mutations from a list of any human SNP. The UMH-HTS resource contains data for all transcripts extracted from Ensembl (HG18-ensembl 54 release). The predictions are based on a combinatorial approach that takes into account the location of the variation at the protein level, that is, in which domain and whether the amino acid is involved in a structural or biological function; it checks for the degree of conservation (data from SIFT); estimates the differences in biochemical properties between the WT and the substituted amino acid (data from BLOSUM62 and Yu’s Biochemical matrix) and finally,as it is well known that missense mutations can have an effect on mRNA splicing, it checks for a potential impact of the substitutions on splicing signals (donor and acceptor splice sites as well as ESE and ESS). The user can use UMD-HTS either to analyze a list of SNPs or to directly access predictions for a given gene or a genomic region.