Hi everyone, I have a methodological question regarding differential analysis of ChIP-seq data. From what I understand, the general workflow is to first define peak regions using MACS3, and then assess differential acetylation using DiffBind. In this case, DiffBind counts the reads using both the .broadPeak or .narrowPeak files predicted by MACS, as well as the .bam files containing the aligned reads, and performs the counting. DiffBind also uses the input files to establish a baseline level of acetylation.
Now, in my case, I have a list of regions for which I need to determine whether they are differentially acetylated. How could I approach this? For example, should I check whether my regions fall within the peaks predicted by MACS and then perform a differential analysis with DESeq2? Or could I extrapolate the results obtained from DiffBind and extract the predicted acetylation levels from DiffBind for my regions of interest? Do these regions need to be 100% overlapping?
In summary, how can I establish differential acetylation in regions that are not those predicted by MACS3? I mention this because I have seen some studies that take the ChIP-seq .bam data and perform DESeq2 directly, but that approach does not consider the input files or the MACS predicted peaks.
I’d really appreciate any suggestions, references, or opinions you can share on this — thank you very much!
For example, in this image, the red area represents the differentially acetylated region returned by DiffBind. The green area is a promoter, the grey tracks correspond to the BAM files (coverage), and the blue ones are the peaks predicted by MACS. If I’m interested in whether the promoter is differentially acetylated, should I say yes, because it falls within the red region? And approximately how much? Should I take the value from DiffBind?