Question: MNase ChIP-seq paired-end data analysis ?
gravatar for ghahhari.nm
10 weeks ago by
ghahhari.nm0 wrote:

Hello everyone! Sorry for long explanation!

I did paired-end ChIP-seq for my input and IP samples and I used MNase instead of sonication for DNA fragmentation. I have duplicate of each sample. After converting from fastq. to sam files, I merged duplicates of each sample with bamtools and also sorted bam files (just to compress files). Then I imported these bam files to galaxy and tried to perform peak calling with the default setting using MACS2 callpeak. I got 82000 regions from narrowpeaks but based on the transcription factor binding sites that I am looking for, I should not have more than 35000 binding sites. I tried to do change settings for MACS2 callpeak as follows and I got 62000 regions:

Band width for picking regions to compute fragment size 500 Set lower mfold bound 150 Set upper mfold bound 1000 Set extension size 100

So the question is if this method of analysis is ok for my special samples regarding using MNase and also a paired-end seq? Do you have any suggestion to optimise the analysis?

Thanks and hope to have your suggestions !

ADD COMMENTlink modified 10 weeks ago • written 10 weeks ago by ghahhari.nm0
gravatar for Jennifer Hillman Jackson
10 weeks ago by
United States
Jennifer Hillman Jackson23k wrote:


There is some prior Q&A about the topic here:

And this is some general advice from our team:

Instead of worrying about the absolute number of peaks MACS2 is [producing], instead investigate by sorting peaks based on enrichment or q-value. Alternatively, the FDR/q-value threshold could be reset to something more restrictive (default q-value = 0.05, maybe change to 0.01).

This publication may also be helpful:

Thanks, Jen, Galaxy team

ADD COMMENTlink written 10 weeks ago by Jennifer Hillman Jackson23k
gravatar for ghahhari.nm
10 weeks ago by
ghahhari.nm0 wrote:

Thank you Jennifer!

Indeed I tried to set the q value which helped a lot and I found a reasonable number of binding events which is different between treatments and it's good! I also tried to follow the Galaxy tutorial by finding the d value first and after peak calling I found very similar results.

I hope you can also help me for my other question now! As I mentioned before we also did same chip-seq of same transcription factor but using sonication. I am interested in comparing the methods to see if we can shift from sonication to Mnase. The question is how can I compare the results? Should I apply the same optimisation for Mnase analysis for sonication data? I did this but the number of binding sites didn't change for sonication Do u think I should first pick the the most relevant binding sites n then compare? Could you please tell me how I can compare the results

Thanks a lot!

ADD COMMENTlink written 10 weeks ago by ghahhari.nm0
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 81 users visited in the last hour