Question: Weird Peak calling results
0
gravatar for parksuhong
24 months ago by
parksuhong0
parksuhong0 wrote:

Hi Everyone,

I need help with Galaxy-Macs' Peak Calling tool.

Using my ChIP-Seq data (Experimental and input), I mapped the reads with bowtie and did the peak calling with Macs. But something doesn't look right. First of all, according the 'Peaks:interval' file, there is only 3000~ peaks that were picked up with all of their FDR equals 100%. And the largest fold enrichment was only 17X. Also, one of the target with the highest fold enrichment from the previous ChIP-Seq (this is my second ChIP-seq repeat) wasn't there in my peaks:interval file. Second, I knew these won't right, so I took a look at it in UCSC genome browser after converting it to BigWig file. There, I looked at that target gene with the highest fold enrichment from the previous ChIP-Seq and there was a robust peak with fold enrichment greater than 70. Why doesn't my peaks:interval file contain target genes with significant fold enrichment like this and just contain 3000~ genes with 100% FDR and lower fold enrichment? 

To me, it looks like something went wrong with the Peaks:interval file during my Peak calling analysis, while files such as peaks;bed file to visualize it under genome browser is perfectly fine. I repeated several times but all I get is the same weird results.

Could anyone please help/guide/instruct/tip me with what could possibly going on with my peak calling analysis?

 

Thank you!

ADD COMMENTlink modified 24 months ago by Jennifer Hillman Jackson23k • written 24 months ago by parksuhong0
0
gravatar for Jennifer Hillman Jackson
24 months ago by
United States
Jennifer Hillman Jackson23k wrote:

Hello,

Without knowing the full parameters and data of the before and after analysis, this is a bit difficult to answer. If you want feedback, replicate this at http://usegalaxy.org (if not already), send an email to galaxy-bugs@lists.galaxyproject.org with links to the shared histories (expected results + unexpected results). Be sure to leave datasets undeleted going all the way back to the input fastq files. If data analysis is missing or deleted, we won't be able to help troubleshoot as well.

That mailing list is private, so your data there will be private as well. Please include a link to this post so we can link the two and include your Galaxy account email (or send from there) in case we need to dig a bit deeper.

Thanks! Jen, Galaxy team

ADD COMMENTlink modified 24 months ago • written 24 months ago by Jennifer Hillman Jackson23k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 84 users visited in the last hour