Leveraging expression quantitative trait loci information in single-cell resolution to identify cell-specific genes for gestational diabetes

Abstract

Background Gestational diabetes mellitus (GDM) is a common pregnancy complication with long-term metabolic consequences for both mother and offspring. While genome-wide association studies (GWAS) have identified risk loci, the cell-type-specific genetic architecture remains poorly characterized due to the limitations of bulk-tissue analyses. Methods We integrated GWAS summary statistics from FinnGen (12,332 cases, 131,109 controls) with single-cell expression quantitative trait loci (sc-eQTL) data from 12 immune cell types in the OneK1K cohort. Using the OTTERS framework and ACAT-O omnibus testing, we performed single-cell transcriptome-wide association studies (scTWAS) to identify GDM-associated genes. We performed bulk-level TWAS as a sensitivity analysis. To investigate the function of significant genes, we performed functional enrichment. Results We detected 14 unique genes significantly associated with GDM across immune cell types (FDR < 0.05), with the strongest signals in CD4 + T cells and monocytes. Notably, ERAP1 and ERAP2 showed associations in 10 of 12 cell types (ACAT. P  = 2.17 × 10 −5 ), and RIOK2 emerged as a shared regulator. Gene Ontology analysis consistently highlighted “antigen processing and presentation via MHC class I” as the top enriched pathway (FDR < 10 −5 ). In contrast, traditional bulk TWAS using GTEx whole blood identified only two genes with nominal associations. Conclusion Our sc-TWAS identifies cell-type-specific GDM-associated genes that are not detected in bulk tissue, highlighting dysregulated antigen presentation in peripheral immune cells. Colocalization prioritizes monocytic LNPEP as the primary causal candidate, while associations for ERAP1 , ERAP2 , and RIOK2 likely reflect complex LD or tissue-specific effects.