MACS2 peak calling and drug treatment

Question

[qudrat.nii]

Hello, I have called peak for a transcription factor (say TF1) using MACS2 on a cell line with DMSO treated with two replicates and the same cell line was treated with a drug that is supposed to degrade the TF1. In DMSO treated number of peak called was 15500 and in drug treated number of peaks reduced to 5000. My question is how to find the identity lost peaks ?

[seidel]

Are you saying there are two replicates of DMSO treatment, and two replicates of Drug treatment? Or two samples, one with DMSO, the other with drug? If the former, you'll have to incorporate a strategy of defining a reference peak set from the replicates.

Answer 1

[seidel]

There are many ways to do this depending on your preferred environment. The GenomicRanges library in R is a good way to interact with this kind of data. However, if you have a bed file of DMSO peaks, and a bed file of drug treatment peaks, and you want to find DMSO peaks NOT overlapped by a drug treatment peak, you could use bedtools intersect:

bedtools intersect -v -a DMSO.bed -b drug.bed

If you have replicates of either treatment, you'll need a strategy for reducing them to a reference set (i.e. keeping only peaks observed in both replicates, and then applying the strategy above, etc.).

[qudrat.nii]

Thank you very much