Mutation severity spectrum of rare alleles in the human genome is predictive of disease type

摘要: The human genome harbors a variety of genetic variations. Single-nucleotide changes that alter amino acids in protein-coding regions are one the major causes phenotypic variation and diseases. These single-amino acid variations (SAVs) routinely found whole exome sequencing. Evaluating functional impact such genomic alterations is crucial for diagnosis disorders. We developed DeepSAV, deep-learning convolutional neural network to differentiate disease-causing benign SAVs based on protein sequence, structural properties. Our method outperforms most stand-alone programs, version incorporating population gene-level information (DeepSAV+PG) has similar predictive power as some best available. transformed DeepSAV scores rare into quantity termed "mutation severity measure" each gene. It reflects gene's tolerance deleterious missense mutations serves useful tool study gene-disease associations. Genes implicated cancer, autism, viral interaction by this measure intolerant mutations, while genes associated with number other diseases scored tolerant. Among known disease-associated genes, those mutation-intolerant likely function development signal transduction pathways, mutation-tolerant tend encode metabolic mitochondrial proteins.