mESC MeRIP-seq sites generated from GSE151028

[hengjwj]

Hi, I downloaded the respective FASTQ (SRR14765640-14765647) from Zhang et al. raw MeRIP-seq data (GSE151028) and generated the list of m6A sites using HISAT2 and MACS2 as described in the methods section. After taking the union of peaks from WT and substracting the peaks from KO, I got 28,083 sites. Of these, only 827 sites have the DRACH motif when in the paper it states that 18,069 RRACH sites were found (paragraph 2 of results section). I expected that DRACH would yield more sites than RRACH so this is really odd. Could you share more details regarding the filtering so I can get closer to the groundtruth your team got? Thank you!

[xieyy46]

Hi,
We apologize for the delayed reply; we've been busy with other projects. First, we took the union of two replicates peaks for WT (using bedtools merge) and obtained 21,356 WT peaks. Then, for KO, we took the intersection of two replicates peaks (using bedtools intersect) and obtained 8,979 KO peaks. Next, we subtracted KO peaks from WT peaks to obtain 18,208 high-confident peaks. After that, I extracted the sequences of these peaks (using bedtools getfasta). If a peak's sequence contained RRACH motifs, we retained it, resulting in 18,069 peaks in the end.

Note that MeRIP-seq got peaks results but not sites results.