About the resource

Type of resource: Variant prediction system
License/availability: Open source
Website: Click here to view website
User documentation:
Relevant publication: Click here to view a publication on this resource


Whole Exome Sequencing is increasingly applied to research and clinical diagnosis of human Mendelian inherited diseases. It typically results in a very large amount of genetic variations ranging from 50 to 90,000 per individual. Depending on the mode of inheritance, only one or two correspond to pathogenic mutations responsible for the disease and present in affected individuals. Distinguishing neutral sequence variations from those responsible for the phenotype is of major interest in human genetics. RD-Connect partner Aix-Marseille University Medical School has developed a new computational combinatorial system UMD-Predictor to efficiently annotate cDNA substitutions of all human transcripts for their potential pathogenicity. This system aggregates information at the nucleotide and protein levels as well as unique features to predict the impact of cDNA mutations at the mRNA level. The accuracy and performances of this innovative approach were assessed using the largest reference variation datasets including more than 140,000 annotated variations (Varibench with dbSNP, Uniprot, Clinvar and PredictSNP) against the seven most used and reliable prediction tools (SIFT, Polyphen, Provean, Mutation Assessor, CONDEL, MutationTaster and CADD). UMD-Predictor consistently demonstrated better results for all parameters. In addition, it is the fastest system performing a full exome annotation in 45s and resulting in the shortest list of candidate mutations, therefore reducing the downstream filtration and validation processes by 64%. Webservices allow its implementation in any bioinformatics pipeline for both exomes and genomes annotation. It could therefore benefit to a wide range of users and applications extending from basic research and gene identification to large population studies and clinical diagnosis.