SC-VAR: a computational tool for interpreting polygenic disease risks using single-cell epigenomic data

MOTIVATION: One major challenge of interpreting variants from genome-wide association studies (GWAS) of complex traits or diseases is how to efficiently annotate noncoding variants. These variants influence gene expression by disrupting cis-regulatory elements (CREs), whose spatial and cell-type specificity are not adequately captured by conventional tools like multi-marker analysis of genomic annotation. Current methods either rely on linear proximity to genes or quantitative trait locus (QTL) data yet fail to integrate single-cell epigenomic information for a comprehensive annotation. RESULTS: We present SC-VAR, a novel computational tool designed to enhance the interpretation of disease-associated risks from GWAS using single-cell epigenomic data. SC-VAR leverages single-cell epigenomic data to predict functional outcomes including risk genes, pathways, and cell types for both coding and noncoding disease-associated variants. We demonstrate that SC-VAR outperforms state-of-the-art methods by predicting more validated disease-related genes and pathways for multiple diseases. Additionally, SC-VAR identifies cell types that are susceptible to disease, along with their specific CREs and target genes linked to risk. By capturing a broad range of disease risks across human tissues at distinct developmental stages, SC-VAR could enhance our understanding of disease mechanisms in complex tissues across different life stages.