Expression based polygenic scores - A gene network perspective to capture individual differences in biological processes

Abstract

Incorporating functional aspects into polygenic scores may accelerate early diagnosis and the discovery of therapeutic targets. Yet, existing polygenic scores summarize information from genome wide statistical associations between SNPs and phenotypes. We developed the novel biologically informed, expression-based polygenic scores (ePRS or ePGS). The method characterizes tissue specific gene co-expression networks from genome-wide RNA sequencing data and incorporates this information into polygenic scores. Performance and characteristics of the ePGS were compared to traditional polygenic risk score (PRS). We observed that ePGS differs from PRS for aggregating information on; i. the relation between different genes (co-expression); ii. the levels of tissue-specific gene expression; iii. the genetic variation of the target sample; iv. the tissue-specific effect size of the association between genotyping and gene expression; v. the portability across different ancestries. Variations in the ePGS represent individual variations in the expression of a tissue-specific gene co-expression network, and this methodology may profoundly influence the way we study human disease biology.