Fig. 5: Prediction of functional transcriptional regulators for Alzheimer’s disease by using post-GWAS analysis. | Nature Communications

Fig. 5: Prediction of functional transcriptional regulators for Alzheimer’s disease by using post-GWAS analysis.

From: TRAPT: a multi-stage fused deep learning framework for predicting transcriptional regulators based on large-scale epigenomic data

Fig. 5

a Workflow of the analysis of data on Alzheimer’s disease. b Venn diagram showing the number of causal SNPs bound near the predicted top and tail TRs. c Scatter plot displaying casual variants bound by the top-ranking TRs, with the sizes of the points representing the magnitudes of their FINEMAP scores. d Bar chart displaying the results of co-localization analysis. Yellow bars represent the number of TRs bound to important causal variants in fine mapping, while gray bars represent TRs bound to background variants. We selected the top and bottom 25 TRs for demonstration. The enrichment line plot represents the binding-induced enrichment of these TRs. Causal variants tended to co-occur with TRs predicted by TRAPT. e Manhattan plot showing the top-ranking causal variant rs10119 obtained from fine mapping. We used the important gene sets analyzed by using the MAGMA software as the input to TRAPT. The bottom tracks represent the peaks of binding of the TRs, where EGR1, RELA, REST, and STAT1 were predicted to rank in the top 10, while the other TRs were predicted to be ranked in the top 25. The bottom 25 TRs showed no binding near rs10119. P-values are calculated by computing the distribution that is greater than the observed ES. f Genome browser displaying the interactions of chromatin, relationships of eQTLs, top-ranking bound TRs, and the epigenetic landscapes of H3K27ac and AD obtained by TRAPT. P-values are calculated by the two-sided Likelihood Ratio test with adjustments.

Back to article page