To increase the discovery power of the analysis conducted in the Yale-Penn cohort, we used the single-variant results obtained from the StructLMM interaction and association tests to conduct genome-wide gene-based analyses considering interactive and main effects, respectively. We applied the Multi-marker Analysis of GenoMic Annotation (MAGMA) gene-based approach28 and a Bonferroni multiple testing correction. Gene-based tests are generally more powerful than single-variant association analysis28, because they combine single-variant signals within genic regions reducing the multiple testing correction burden. We performed a functional annotation of the variants identified using data from combined annotation dependent depletion (CADD)29, RegulomeDB30, and 15-core chromatin state information across multiple brain tissues. Using genotype-tissue expression (GTEx) V831, we tested the effect of the variants identified on the tissue-specific transcriptomic profiles of the surrounding genes (±1 Mb of the gene transcription starting site), considering a false discovery rate at 5% for the genome-wide multiple testing correction. To investigate the loci identified further, we performed single-variant and gene-based phenome-wide scans leveraging the GWAS Atlas (available at https://atlas.ctglab.nl/)32.
