Computing the significance of overlapping ChIP-seq peaks

Question

[Marcel]

Hello everyone,

I have ChIP-seq data of stalled Pol2 (13.000 peaks) and a protein that mostly binds in promoter regions (250 peaks). Now I want to know whether there is a significant overlap between the two.

What I did try so far, is using "bedtools fisher" with the two datasets. However, as far as I understand, this will not take care of the fact, that both datasets mostly bind in promoter regions. Of course, it's telling me that the overlap is extremely significant.

In other words: I want to know whether my protein of interest binds significantly more often than expected to a region where Pol2 is sitting, considering that Pol2 already occupies a large fraction of promoters with its 13.000 peaks.

Any ideas on how to do this properly?

Thanks.

[igor]

Have you checked these previous threads?

• How Do You Calculate If Two Sets Of Genomic Regions Overlap Significantly?
• Association between bed files - statistical significance

[Marcel]

Sorry for the late reply. I've had a look at these threats and they mention IntervalStats (which was also suggested in the other reply to my thread) which appears to fit my purpose.. however, I can't get it running. Other than that, I'll have a look at Cooccur, which was also mentioned. Thanks for your reply!

Answer 1

[colin.kern]

There's a tool called IntervalStats that does this. Here is their paper: https://academic.oup.com/bioinformatics/article/28/5/607/247792

[Marcel]

Sorry for the late reply and thanks for the suggestion. This tool indeed seems to be doing exactly what I want. However, it's throwing me errors saying "Malformed BED line".. but my .bed files are allright. I've been using them for other analyses already.