How different the different ChIP-seq samples are ....

Question

[Dataminer]

Hi!

I have two samples from patient A and B. On these samples ChIP-seq was performed as untreated, treated with D1, treated with D2 and treated with D3.

In the end I have four ChIP-seq profiles for each patient, like: A_un, A_D1, A_D2, A_D3, B_un, B_D1, B_D2 and B_D3.

What I want to look at is, if these two samples show some kind of relatedness either before or after treatments.

What I am doing is calling peaks for these untreated and treated ChIP-seq samples and clustering them and looking at the samples if they are appearing in same cluster or far away clusters.

I was just thinking if I could do some kind of data pre processing to remove any redundant peaks (if necessary) from the samples as the two samples are from diabetic patients (same disease). Or is their any better way to get the difference or the relatedness between the samples?

Moreover can mmseq can be applied to a ChIP-seq data like the one above?

Could you please share your knowledge and information?

Thank you

[Bharat Iyengar]

I think you are aware of IDR because you have a post on that. IDR can be used to remove these peaks.

Answer 1

[Ying W]

Do you mean this MMSEQ?

I think the simple approach would be to call peaks for each and compare A vs B for replicate consistency then compare _un vs _D1/_D2/_D3

You can use IDR to call peaks since you have replicates. A quick way to visualize would be to make a 4-way venn diagram of shared vs unique peaks

If you are further interested in relatedness between samples, you can calculate correlations or create heatmaps (DiffBind will do this for you)