Linking Genetic Variants to Complex Traits with Single-cell Genomics and RNA Modification

Genome-wide association studies have extensive success on identifying genetic variants associated with complex traits and diseases. However, the underlying molecular mechanisms are mostly unknown. Though many efforts on linking variants to functions have been through functional fine-mapping, colocation of expression QTLs with GWAS, etc., there are limitations in each approach. We aimed to fill the gaps by taking two emerging directions in the field - single-cell genomics and RNA modification. For the first of direction, we profiled single-cell multiomics on lung lymphocytes and integrated them with GWAS to identify asthma risk genes and their cell-type specific functions. We were able to identify distinct transcriptomes for lungs by cell type, but the differences in chromatin accessibility were subtle. We further identified open chromatin regions (OCRs) in each lung immune cell type and showed their unique contributions to asthma heritability beyond commonly used OCRs in blood immune cells. Using the lung OCRs and previously fine-mapped variants for COA and AOA, we identified 43 cis-regulatory elements (CREs) likely contributing to asthma risk. We also mapped target genes of 34 of these CREs and found they collectively showed evidence of being disease causing, based on gene functions. We highlighted two genes, CCR4 and LRRC32 with their supporting variants that show enhancer activity in specific lung immune cell types. Lastly, we built cell-type level gene regulatory networks (GRNs), which shed light on the regulation of gene expression in disease through several asthma risk genes that are TFs. Our results demonstrate the utility of single-cell multiomics in interpretating GWAS risk loci. For the second direction, we investigated the role of one type of mRNA modification – N6-adenosine (m6A) in human genetics. We curated m6A peaks from various cell lines and tissues and systematically assessed their contribution to the heritability of a wide range of complex traits and diseases. Overall, we observed broad enrichment of m6A peaks in trait heritability. To identify putative causal variants with potential m6A-mediated roles, we leveraged m6A peaks to annotate published fine-mapped data. The resulting high-confident SNPs in m6A peaks were mostly non-coding and their likely downstream effects were on RNA processing. Lastly, we leveraged published m6A-QTL datasets to nominate risk genes for immune-related traits. We discovered 4 to 13 genes per trait, and they together were strongly enriched for type I interferon signaling. All these results together thus supported mediating roles of m6A to diverse traits and diseases.