paperKB
coga / coga-kb
Help
Sign in

Chunk #47 — Methods — Identification of DMRs

Source
Hotspots of aberrant epigenomic reprogramming in human induced pluripotent stem cells.
Embedded
yes

Text

Non-CG mega-DMRs (Fig. 5) were identified by comparing H1 to ADS-iPSC mCHG and mCHH smoothed methylation profiles. The average methylation level of mC called (1% FDR) in the mCHG and mCHH sequence context was determined in 5-kb windows (sW). The genome was scanned considering groups of 10 adjacent windows sW over a distance less than 50 kb. The set of 10 smoothed values for mCHG in the H1 sample was compared to the set of set of 10 smoothed values in the ADS-iPSC sample using the Wilcoxon test. For both sets, at least 4 non-missing data points (that is, with sequence coverage) were required. Resulting P values were corrected with the Benjamini-Hochberg method. Regions with P value < 0.01 (1% FDR) and 8-fold enrichment of methylation level were identified, and regions closer than 100 bp were joined. This was repeated for the mC in the CHH sequence context. Lastly, mCHG and mCHH DMRs overlapping or closer than 100 kb were joined and the final set of regions was checked for having mCHG+mCHH fold enrichment of at least 2-fold between H1