TWAS and SMR methods can also be applied to protein quantitative trait loci (pQTL) datasets, inferring whether genetic variation associated with MDD confer altered protein levels. Recently, pQTL data from the dorsolateral prefrontal cortex (DLPFC) has been prepared to perform proteome-wide association study (PWAS),86 using genotype-protein data from two datasets, referred to as ROSMAP and Banner et al. We followed the same procedure as the study originally performing PWAS, which included performing PWAS using both ROSMAP and Banner et al. panels, treating the larger ROSMAP panel as the discovery sample, and the Banner et al. panel as a replication sample. Proteins were identified as statistically significant in ROSMAP if pFDR < 0.05 (correcting for all proteins tested) and considered replicated in Banner et al. if pFDR < 0.05 (correcting for number of proteins tested for replication). PWAS was performed using FUSION software with the in-built downstream colocalization analysis using coloc (PP4 > 0.8). As in the original PWAS, we used the HEIDI test within SMR to confirm evidence of colocalization based on the ROSMAP dataset (HEIDI p > 0.05).
