Abstract
Liver diseases encompass a wide range of etiologies and involve highly heterogeneous cellular environments, yet the specific cellular states through which genetic risk contributes to disease remain incompletely understood. In this study, we integrated genome-wide association study (GWAS) data from six liver diseases and two metabolic traits with transcriptomic profiles of approximately 168,000 human liver cells at single-cell resolution, using the single-cell disease relevance score (scDRS) approach. Our results revealed that disease-associated genetic signals are predominantly localized to non-parenchymal cells-particularly liver sinusoidal endothelial cells (LSECs), cholangiocytes, and specific subsets of lymphocytes. Notably, we identified marked intra-cell-type heterogeneity, with disease associations confined to specific subpopulations exhibiting immune activation or stress-responsive transcriptional programs. For example, autoimmune and viral liver diseases were linked to immunologically active LSECs and cholangiocytes, whereas their metabolically active counterparts showed no enrichment. These findings highlight the necessity of resolving liver cell complexity to uncover the functional basis of genetic risk and suggest that susceptibility to liver disease is driven by specialized cell states within broader cellular categories. Our study provides a refined cellular map of liver disease susceptibility, offering new perspectives for understanding pathogenic mechanisms and informing targeted therapeutic strategies.