Skip to main content
. 2025 Jul 3;5(9):100929. doi: 10.1016/j.xgen.2025.100929

Figure 4.

Figure 4

MSA reveals the role of 5hmC in human tissue identity definition

(A) t-SNE plot of bulk tissues clustered by 5hmC profiles.

(B) Heatmap of representative one-vs.-rest 5hmC tissue signatures (rows) across tissues (columns). The annotation bars are colored by tissue type.

(C) Dot plot showing enrichment of 5modC tissue signatures in 5hmC signatures. Circle sizes represent −log10(FDR-adjusted p values) from one-tailed Fisher’s test.

(D) Heatmap showing enrichment of genes linked to hyper-5hmC and hypo-5modC CpGs (columns) in tissue-specific expression gene sets identified using GTEx data of matched tissue types (rows). The annotation bar represents whether the query gene set is hyper-5hmC linked (yellow), hypo-5hmC linked (blue), or both. A one-tailed Fisher’s test for enrichment was used.

(E) Grouped bar chart showing the number of tissue-specific genes marked by no tissue-specific cytosine modifications (gray), hypo-5modC (blue), hyper-5hmC (yellow), and both modifications (green) for each tissue type.

(F) Tissue-biased genes marked only by tissue-specific 5hmC. The x axis shows mean 5hmC over linked probes; genes are colored by tissue specificity.

(G) Violin plots of Pearson correlations between tissue-specific methylations and expression of linked genes, grouped by tissue and the two types of modification changes.