DMR analysis in SeqMonk

biodobe · 07-04-2019, 10:14 AM

I'm running DMR analysis in SeqMonk using context specific methylation extraction files produced in Bismark. I've created probes using the read position probe generator and have quantified using the Bisulfite Methylation Quantitation Pipeline.

When I run the Logistic Regression statistical test for replicate data, many areas of interest are identified, which I can then scroll and see some obvious DMRs of interest using the track display. By obvious, I mean there are regions which are highly methylated in one replicate group (value >50) that lack any methylation in the other group. This is not due to low coverage; I have achieved 80x genome coverage.

However, when I use the "Filter on Value Differences --> Individual probes" to identify DMRs of >= 20% difference in methylation, no probes are returned.

Can anyone suggest why these regions are not being picked up by the filter?

**I do not have this problem when I use coverage files from Bismark. However, I don't have the methylation by sequence context information if I rely on the .cov files for this analysis, which is why I'm using the files produced by the methylation extraction step in Bismark.

simonandrews · 07-05-2019, 04:00 AM

You say you don't get the same thing when you use the .cov files from bismark? That sounds like something fundamentally isn't right in your data import then, because you should get identical data sets from the cov and the .txt.gz bismark files once you've merged together the top and bottom strands.

Could you make up a vistory in seqmonk to show what you've done with the data and the analysis you've run along with some views of regions you think are obviously being missed and either mail it directly to me (my email is on the Babraham Bioinformatics web page) or file it as an issue on the seqmonk github so I can take a look at what you have.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Background for GO term enrichment analysis of RRBS DMR data?	pig_raffles	Bioinformatics	0	03-05-2019 01:18 PM
Issue with BiSeq for DMR analysis	shocker8786	Bioinformatics	13	10-12-2016 05:41 PM
How to define a region while conducting BS-seq DMR analysis	zinky	Bioinformatics	0	12-11-2012 06:36 PM
SeqMonk: Export features (e.g. CpG Islands) from Ensembl for import into SeqMonk?	jjw14	Bioinformatics	8	06-08-2012 12:35 AM

07-04-2019, 10:14 AM	#1
biodobe Junior Member Location: USA Join Date: Apr 2018 Posts: 3	DMR analysis in SeqMonk I'm running DMR analysis in SeqMonk using context specific methylation extraction files produced in Bismark. I've created probes using the read position probe generator and have quantified using the Bisulfite Methylation Quantitation Pipeline. When I run the Logistic Regression statistical test for replicate data, many areas of interest are identified, which I can then scroll and see some obvious DMRs of interest using the track display. By obvious, I mean there are regions which are highly methylated in one replicate group (value >50) that lack any methylation in the other group. This is not due to low coverage; I have achieved 80x genome coverage. However, when I use the "Filter on Value Differences --> Individual probes" to identify DMRs of >= 20% difference in methylation, no probes are returned. Can anyone suggest why these regions are not being picked up by the filter? *I do not have this problem when I use coverage files from Bismark. However, I don't have the methylation by sequence context information if I rely on the .cov files for this analysis, which is why I'm using the files produced by the methylation extraction step in Bismark. Last edited by biodobe; 07-04-2019 at 11:16 AM.*

07-05-2019, 04:00 AM	#2
simonandrews Simon Andrews Location: Babraham Inst, Cambridge, UK Join Date: May 2009 Posts: 871	You say you don't get the same thing when you use the .cov files from bismark? That sounds like something fundamentally isn't right in your data import then, because you should get identical data sets from the cov and the .txt.gz bismark files once you've merged together the top and bottom strands. Could you make up a vistory in seqmonk to show what you've done with the data and the analysis you've run along with some views of regions you think are obviously being missed and either mail it directly to me (my email is on the Babraham Bioinformatics web page) or file it as an issue on the seqmonk github so I can take a look at what you have.