Atlas Khan , Department of Biomedical Informatics, Columbia, University, New York, NY, 10032, USA
Kai Wang , Department of Biomedical Informatics, Columbia, University, New York, NY, 10032, USA
Many rare and common genetic variants, including SNPs and CNVs, are reported to be associated with mental disorders, yet more remain to be discovered. However, despite the large amount of high-throughput genomics data, there is a lack of integrative methods to systematically prioritize variants that confer susceptibility to mental disorders in personal genomes. Here, we developed a computational tool: a deep learning based scoring system (ncDeepBrain) to analyze whole genome/exome sequencing data on personal genomes by integrating contributions from coding, non-coding, structural variants, known brain expression quantitative trait locus (eQTLs), and enhancer/promoter peaks from PsychENCODE. The input is whole-genome variants and the output is prioritized list of variants that may be of relevance to the phenotypes. For population studies, our method can help prioritize novel variants that are associated with disease susceptibility; for individual patients, our method can help identify variants with major effect sizes for mental disorders.