Looking only at m-C in DMR pair

[EstabilloLL]

Hi,

I am running dmr pair for differential methylation scoring. My bed files are bgzipped and indexed.

Here is the code used: ./modkit dmr pair -a CTRL.wf_mods.bedmethyl.gz -b DIFF.wf_mods.bedmethyl.gz -o output.bed -r cpg_islands_ucsc_cleaned.bed --ref GRCh38.p13_genomic_chr_only_plus_mt.fa --base C --assign-code m:C --log-filepath dmr.log

I only want to look at 5mC modifications and not 5hmC modifications. I have tried specifying with --assign-code m:C, but the output still has h and m - and thus calculates the scores based on both.

I have also tried filtering out only the lines with "m" in the 4th column in the bed files, but the command won't work and shows "invalid bedMethyl records"

Is there any way to only look at 5mC modifications and disregard 5hmC completely in dmr pair?

Thank you very much for the help.

Best Lance

[ArtRand]

Hello @EstabilloLL,

Is there any way to only look at 5mC modifications and disregard 5hmC completely in dmr pair?

The best way to do this is to use --ignore h whilst generating the pileups. For example

$ modkit pileup ${bam} - --ignore h | bgzip -c > pileup_ignoreh.bed.gz

I have also tried filtering out only the lines with "m" in the 4th column in the bed files, but the command won't work and shows "invalid bedMethyl records"

There is an internal check that the records haven't been filtered this way so I would expect that you get this error.