Leveraging genomic and transcriptomic data of diverse ancestry to uncover mechanisms of psychiatric risk in the adult and developing brain.

We explore strategies to harness ancestral diversity in PsychENCODE Consortium Genotype-Expression (GEx) reference panels (adult and developing brain) and Psychiatric Genomics Consortium GWAS data to improve genetically regulated expression (GReX) models and their use for Transcriptome-wide association study (TWAS) discoveries, uncovering previously unknown aspects of psychiatric functional genomics. We trained multiple GReX models on rigorously constructed GEx panel subsets, generated by downsampling, segregating, and/or mixing samples of Admixed African and European ancestries, and based on disease status. Ancestry-specific GReX genes were enriched in pathways involving mitochondrial functions, organelle structure, and metabolism. These models were integrated with ancestry-specific GWASs to conduct bipolar disorder, major depressive disorder, posttraumatic stress disorder, and schizophrenia TWAS. TWAS signals obtained by applying AA- and EUR-specific GReX models to an ancestry-specific GWAS were largely concordant, and mismatched-TWAS (e.g. AA-GReX applied to EUR-GWAS) revealed biologically meaningful signals missed by matched-TWAS. Shared signals across the four disorders were more prominent in the developing brain, involving genes such as H4C13, ZSCAN12P1, and FLOT1, and pathways related to megakaryocyte and muscle development, and neurotransmitter regulation. Overall, we demonstrate concordance in shared TWAS signals across GReX models and provide insight into GReX-specific detectable genes and pathways.