Fig. 3: Candidate gene summary.

Depicted is a summary of study results. We used a deep learning algorithm to perform a GWAS of CMR-derived LVMI in 43,230 individuals, finding 12 independent loci associated with LVMI. Using proximity to lead variants, expression quantitative trait locus (eQTL) analysis, transcriptome-wide association studies (TWAS), Hi-C analysis, LV tissue-specific expression levels, and prior evidence, we identified candidate genes across the 12 loci. Candidate genes were enriched for genes involved in stress response and neurohormonal regulation, cardiac structure and cardiomyopathy, and cell signaling/function (gray box).