comparison between two samples in chip-seq

[emilyjia2000]

Hello,

I would like to do the comparison between two samples in chip-seq data, to see the similarity or difference between these two samples. I used MACS to detect the peak calling, I don't know how to do the comparison, any input would be appreciated.

Thanks,

[emilyjia2000]

I might not elucidate the fact clearly. After the peak calling, I could get the overlaps between the peak regions between two different samples (from different cell lines), but this could not demonstrate the similarity or difference statistically.

Does anybody know any tools that could help on solving this?
Is there any way to check the correlation between two samples?

Thanks,

[vanbug]

Hi,
If they are similar samples for the same TF or from same cells, the comparison is possible. The first step is to normalize them, by having equal number of reads in both of them. This can be done by sampling technique, just get the lower number among the two, and subtract the number of extra reads randomly from the other sample, which equalizes them.

Then, you can have the peak comparisons at some gene cadidates, for a extensive search you have to get the gene list where the binding profiles are observed and have a correlation analysis, this will answer few questions.

Also, you can overlay the H3K4(active) and H3K27(repressive) signals to determine the state of chromatin.
These is one kind of analysis you can do

Cheers

[danielr]

I'd take the equally-sized peaks from one and count reads from the other, and vice versa. And compare to random regions.

seqminer is also nice for this problem, as it sometimes manages to separate clusters of overlapping and non-overlapping sites.

Equalizing sequence depth doesn't really help, as the greatest factor for sensitivity is signal to genomic background ratio, and that's different for every ChIP, even with the same antibody.