modkit dmr results- question

[khushimamgain]

Dear authors,
Thankyou for making ModKit, it is a really nice tool. I have some questions on the interpretation of the results I got from the DMR pair between two conditions after m6A profiling and I would be grateful to you if you could explain to me what it means.

What I did:

• Dorado basecalling for m6a and pseudouridine, alignment to transcriptome, followed by modkit pileup to get bed methylfiles
• modkit dmr pair between two conditions with the following command to get results:
  dmr_result=differential_tx_CT20vsPerKO20.bed modkit dmr pair \ -a ${A1} \ -a ${A2} \ -a ${A3} \ -b ${B1} \ -b ${B2} \ -b ${B3} \ --header \ --min-valid-coverage 20 \ -o "$out_dir"/dmr/${dmr_result} \ --base A \ --ref ${ref} \ --base T \ --threads ${threads} \ --log-filepath dmr_2.log
• Filter based on effect size and balanced p-value.

issue:

One of the positions is different significantly on the transcript, ENSMUST00000174287, showing its percentage going from 80% in A to 38% in B, meaning a decreased modification in condition B.

However, in the input pileup bed files (condition A: KM4,5,6, condition B: KM7,8,9), the percentage values is A are, 80, 20, 20 and in B are 66, 40, 38. Based on just the % values, it seems that there only one sample in A has higher % than in B

Question

I am confused about how the differential marks are calculated and if one sample can cause so much effect. If that is the case, could you please tell me how I can interpret this?
Any input from you would be really helpful, as I might have also misunderstood the results.

Thankyou in advance.
Khushi.

[ArtRand]

Hello @khushimamgain,

Could you attach, paste text, or send me the bedmethyl files you're using for $A1, $A2, $A3, $B1, $B2, and $B3. It's hard for me to parse these images.

Thanks. If you decide to email the data, send it to art.rand[at]nanoporetech[dot]com.

[khushimamgain]

Hello,
Thank you for your reply. I have emailed the information pasted as text.
Hoping to hear back from you soon.
Thankyou,
Khushi.