Fig. 5: Integrating eight gene prioritization predictors to identify most likely causal genes.

a, Prioritization of 220 likely causal genes using eight predictors. Blue circles represent the eight predictors used to prioritize causal genes, which are as follows: (1) a gene in the region harbors a variant that ClinVar classifies as having evidence for being pathogenic for a cardiovascular-relevant monogenic disorder (Supplementary Table 34); (2) a gene in the region has been implicated by an effective drug targeting the protein and/or a positive MR study suggesting a causal effect of the protein on CAD (Supplementary Table 31); (3) either of the two top prioritized genes in the region from PoPS (Supplementary Table 24); (4) a gene in the region has an eQTL in a CAD-relevant tissue from GTEx or STARNET for which the lead eSNP is in high linkage disequilibrium (LD) (r² ≥ 0.8) with the CAD sentinel variant (Supplementary Tables 27 and 28); (5) a gene for which a mouse knock-out has a cardiovascular-relevant phenotype (Supplementary Table 35); (6) a gene in the region harbors a protein-altering variant that is in high LD (r² ≥ 0.8) with the CAD sentinel variant (Supplementary Table 31); (7) a gene in the region has been shown to have a rare variant association with CAD in a previous WES or genotyping study (Supplementary Table 31); (8) the nearest gene to the CAD sentinel variant. Numbers in the blue circles indicate, firstly, the number of genes for which this predictor agreed with the most likely causal gene, secondly, the number of genes for which this predictor provided evidence for at least one gene, and in parentheses, the percentage agreement (that is, the first number as a percentage of the second). The central histogram shows the number of agreeing predictors that supported the 220 prioritized genes by the number of genes. b, Predictors for 44 most likely causal genes strongly prioritized by at least four agreeing predictors. The matrix denotes predictors that supported the most likely causal gene (colored red) for each of the 44 most likely causal genes with at least four predictors that supported the gene. Genes are ordered by number of agreeing predictors. The sentinel variant for the association with the smallest P value for CAD is shown for each gene. Full details of the causal gene prioritization evidence for all 279 genome-wide associations are presented in Supplementary Table 31 and the 79 most likely causal genes with three agreeing predictors are displayed in the same format in Supplementary Fig. 1.