Fine-Mapping Genome-Wide Associated Loci using Multi-omics Data to Identify Mechanisms Affecting Serious Mental Illness

Genome-wide association studies have been key for identifying genetic variation associated with psychiatric disorders. Whenever these GWAS are based on large sample sizes, however, they implicate a plethora of single nucleotide polymorphisms (SNPs) in risk. This polygenicity presents challenges for mapping risk variation onto the biological mechanisms that predispose

individuals to illness. Many studies have integrated genomic and transcriptomic variation with the goal of colocalizing the GWAS SNP associations and cis transcriptional patterns determined by expression quantitative trait loci (eQTLs), as well as other QTLs. In some instances, these studies highlight one or more genes whose transcriptomic variation is driven largely by variation in specific

risk SNPs. For a substantial fraction of the risk loci, however, colocalization is inconsistent across studies or no effect on transcription is observed. These missing links between genetic risk variation and biological variation could be due to many factors, including cell-type specificity, developmental

patterns, or missing -omics characterizations. Notably, bulk tissue and even single cell mRNA levels are imperfect predictors of the cellular levels of the proteins they code for. We hypothesize that a substantial portion of these missing links is due to our limited knowledge of how proteomic variation relates to genetic variation in the human brain. SNPs can regulate the proteome via

mechanisms that “skip” transcript levels and protein levels are tightly regulated by posttranslational modifications (PTMs) that are not readily predictable from the transcriptome. We propose to characterize transcriptomic and proteomic variation in human post-mortem brain, specifically protein expression (Aim 1); PTMs (Aim 2); map genetic variation onto

transcriptomic (eQTLs) and proteome and PTM variation (pQTLs and PTMQTLs) and evaluate their interrelationships (Aim 3); and then perform colocalization analysis to inform the biological pathways by which genetic variation confers risk to psychiatric disorders (Aim 4). In our preliminary proteogenomic experiments, we combined proteomics with SNP genotyping to identify pQTLs.

We discovered that a substantial fraction of pQTLs bypass the transcriptome (~50%), in line with another recent human brain pQTL study and our hypothesis. Our aims are consistent with goals from RFA-MH-21-100: (1) develop novel proteomic and other omics resources; (2) use them to map how genetic risk variation influences

omics features in neural tissue and cell types; and (3) provide a high confidence set of causal variants, genes, and isoforms that likely contribute to disease risk, enhancing our insights into proximate disease mechanisms.