Multi-trait genome-wide analysis identified risk loci and candidate drugs for heart failure

Figures

• 

  Figure 1. Overall study design Employing the largest available GWAS summary statistics for HF encompassing 47,309 cases and 930,014 controls, as well as CAD with 181,522 cases and 984,168 controls, we explored the polygenicity and genetic correlation between them using LDSC, followed by MTAG to identify significant SNPs. Functional annotation at the SNP level was performed to identify risk loci paired with MTAG on replication phase to confirm the MTAG-identified risk loci. Additionally, TWAS was conducted to uncover HF risk genes. Subsequently, four gene-based analyses (pathway enrichment, single-cell enrichment, protein-protein interaction network, and animal knockout models) were performed to confirm the role of TWAS risk genes in the development of HF. Finally, drug repurposing was employed to identify potential therapeutic drugs for HF.

• 

  Figure 2. MTAG analysis boosted SNP discovery power for HF Manhattan plots for MTAGHF (A) and GWASHF (B). Horizontal red dashed lines indicate the genome-wide significance threshold (p < 5 × 10−8). SNPs are labeled in place on chromosome coordinates.

• 

  Figure 3. Functional annotations of candidate SNPs (A) Functional annotation of candidate SNPs. (B) Distribution of candidate SNPs across 15 categories of minimum chromatin state (chromatin states with a value ≤7 are considered as open chromatin regions). (C) RegulomeDB scores of candidate SNPs. RegulomeDB scores from 1b to 2c are considered likely to affect transcription factor binding.

• 

  Figure 4. Functional enrichment of significant MTAGHF SNPs A total of 4,899 SNPs were analyzed with GARFIELD. Enrichment by (A) genic regions, (B) tissue-specific DNase I hypersensitive site (DHS) regions, (C) tissue-specific chromatin states, and (D) tissue-specific histone-modified regions. In all plots, enrichment odds ratio were plotted against −log10(p value). The dotted horizontal lines mark the statistical significance threshold of p = 0.05/1005. Size of dot markers (NAnnotThesh) is proportional to the number of independent significant SNPs in the specific annotation category, while in subplots (C) and (D), markers are also color-coded by tissue categories.

• 

  Figure 5. Contextual analysis of TWAS risk genes (A) Bubble plot of 13 enriched pathways. The x axis represents pathway names, and the y axis represents –log10(Padj). Size of the bubbles indicates the number of genes enriched, while the color of the bubbles indicates the levels of the –log10(Padj, i.e., adjusted p value). (B) Enrichment of risk genes at cellular level. Color scale indicates the aggregated risk of the cells with red for high risk and blue for low risk. (C) Protein-protein interaction (PPI) sub-network of 10 hub risk genes. Gene nodes are arranged from left to right based on their degrees, with round rectangle nodes representing the hub genes.