Large-scale sequencing efforts of genomes from patients often lead to the identification of DNA variants of unknown significance (VUS), i.e. of genomic variations that cannot be immediately recognized as pathogenic mutations or benign DNA changes. Despite many in silico predictors exist, none of them takes advantage from the wealth of information contained in the positional clustering of mutations already detected in disease-associated genes.
We have therefore developed MutScore, a new pathogenicity predictor that integrates unsupervised features of single DNA variants with information derived from such clusters. The predictive model for MutScore was trained with a random forest approach on medically-relevant mutations and subsequently tested against various genomic databases for both hereditary conditions and cancer (ClinVar, HGMD, and DoCM), achieving a very high performance.
The use of MutScore on 840 genes from the ClinVar database also allowed the detection of significant clusters of disease-associated and of benign variants in 505 and 345 of them, respectively, revealing protein domains with diverging functional importance. In addition, an open-access web-based application, MutLand, was developed to provide a comprehensive graphical landscape of all known medically-relevant and clearly benign DNA variants for individual genes, as a help in appraising new VUS identified in clinical testing.
Altogether, our work reveals the presence of widespread clustering of missense variants associated with normal and clinical phenotypes and that this information can be systematically used to improve and to understand pathogenicity at the molecular level.
Mathieu Quinodoz
Institute of Molecular and Clinical Ophthalmology Basel (IOB)
Mittlere Strasse 91, 4031 Basel, Switzerland
IOB websiteMutScore and MutLand are freely available for non-commercial use. For other uses, please contact IOB.
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
MutScore scripts can be downloaded on our github page:
github.com/mquinodo/MutScoreMutScore values for all missense variants for hg19 and hg38 genome builds can be downloaded here:
MutScore-v1.0-hg19MutScore values for ANNOVAR annotation can be downloaded here:
MutScore-v1.0-hg19-ANNOVARMutLand for all genes with variants in ClinVar can be downloaded here:
MutLand-ClinVar-20220109The publication can be accessed freely on the AJHG website:
https://doi.org/10.1016/j.ajhg.2022.01.006For MutScore-batch, please visit this webpage:
iob-genetic.shinyapps.io/mutscore-batch