We used a range of methods and functional genomic datasets to gain insight into the associated variants, genes, and pathways that may be dysregulated in MD. These included three rigorous “high-confidence” approaches: SNP-based fine-mapping of MD-associated loci and integration of expression and protein quantitative trait loci (eQTL and pQTL) to infer genetically driven MD case-control differences in RNA and protein expression. These are referred to as transcriptome- and proteome-wide association study approaches (TWAS and PWAS) and were reported when summary data-based Mendelian randomization (SMR), colocalization (COLOC), and expression-based fine-mapping (of eQTLs and pQTLs, in FOCUS) analyses all aligned to indicate a common gene. We also mapped associated loci to genes using standard gene-based association analysis in fastBAT, chromatin interaction datasets (Hi-C), and applied a gene prioritization package, psychiatric omnilocus prioritization score (PsyOPS) (see STAR Methods).
