Quantitative comparison between 2 IP ChIP-Seq samples, each with an input control.

[blancha]

My initial question was how to quantitatively compare peaks from 2 ChIP samples, with accompanying inputs for each sample.
macs2 bdgdiff allows one to take into account the inputs, but none of the other existing programs seem to allow this option.
After doing research, the best option seems to just ignore the inputs, and compare the IP samples directly.

According to the MMDiff authors:
"Input correction is not necessary for differential peak calling as the local biases such as sequencing bias should affect all considered sample in the same way."

[ShaunMahony]

I think that depends on what you are comparing. If you are comparing binding in different cell lines or in significantly different conditions, then you can indeed have sample-specific artifacts and ignoring the inputs will be the wrong thing to do.

Shameless plug: Our MultiGPS software allows you to compare transcription factor binding across conditions in a principled way.

[[email protected]]

One tool performing differential peak calling with replicates and supporting multiple input-DNA is THOR.

Check here http://www.regulatory-genomics.org/THOR

[mia2025]

My favorite tool for this kind of analysis is Diffbind. It requires peaks called for each individual replicate using the respective inputs and then generates a list of peaks which are statistically differentially bound (I believe it uses DESeq for this, but you can also switch to use EdgeR if you prefer). The good news is that the tutorial is very user-friendly and in depth and actually uses ERa ChiP Seq data from different cell types as a tutorial.

Hope this helps

DiffBind R Package : https://bioconductor.org/packages/re.../DiffBind.html