Skip to main content
. 2020 Oct 19;11:5270. doi: 10.1038/s41467-020-18965-w

Fig. 3. Differential 5hmC occupancy analysis in PDAC cfDNA as compared to cfDNA from non-cancer patients.

Fig. 3

a MA-Plot showing all genes with differential 5hmC representation. Red and green, respectively denote increased or decreased 5hmC density in PDAC compared to non-cancer with adjusted p-value <0.01, derived by Benjamini–Hochberg method. b IGV genome browser snapshot of YAP1 locus showing the increased 5hmC signal intensity in PDAC samples compared to non-cancer controls. c GSEA of 794 genes with the most statistically significant differential 5hmC representation (adjusted p-value < 0.01, by Benjamini–Hochberg method) and filtered for fold change in 5hmC representation (|(5hmC-PDAC/5hmC-non-cancer)| ≥ 1.5) and minimum average expression (log2(average representation) ≥ 3.5) in PDAC cfDNA as compared to non-cancer samples. Log10 FDR values are derived from Kolmogorov–Smirnov test. Pathways with represented in genes with increased and decreased 5hmC are denoted with red and green, respectively. d MDS of pancreatic cancer (orange) and non-cancer (blue) cfDNA samples using 11,855 genes with statistically significant (adjusted p-value < 0.01, Benjamini–Hochberg method) increase or decrease in 5hmC. Note reasonable partitioning of PDAC from non-cancer samples. e, f MDS of pancreatic cancer (orange) and non-cancer (blue) cfDNA samples from this study (e) and Song et al. (f) using 794 genes with statistically significant differential 5hmC representation (adjusted p-value < 0.01, Benjamini–Hochberg method) and filtered for fold change in 5hmC representation (|(5hmC-PDAC/5hmC-non-cancer)| ≥ 1.5) and minimum average expression (log2(average representation) ≥ 3.5).