Entering edit mode
Hello everyone,
I need to do differential ChIP-Seq analysis to find the binding of genes between 2 conditions. I know there are lots of tools for that but I decided to use DESeq2 for that. I used MACS for peak calling and I had BAM files. Then used featureCounts to generate counts matrix. To this count matrix, I used DESeq2.
Everything looks fine but I am not sure is this method sensible or not but I needed differential analysis with gene names. Can I do this without a doubt?
Thank you.
I'll insert a shameless plug for csaw here. Once you have your window-based count matrix, you can use any differential testing framework of your choice (including DESeq2) to generate p-values, and then continue on with the analysis pipeline.
I second the plug for csaw.
Sorry for late reply and thank you so much, I will look to csaw package.
But my question is; using only the DESeq2 (used featureCounts to have CountMatrix from with BAM files) is wrong?
Thank you.
Whether or not this approach is "wrong" depends entirely on how you defined the regions in the first place. See my comments in A: Filtering for ATAC-seq.
I am thinkg about my GTF file. The file I used in featureCounts step, includes genes and gene predictions (From USCD) and I used them for counting the MACS results. Is that make a big difference because I am not sure GTF file includes enhancers of genes. This is the point I am not sure about.
It's hard to figure out what you actually did. Using a GTF file to "count the MACS results" makes no sense; I assume that what you meant was that you counted the number of reads assigned to each gene with featureCounts. I'll also assume you're referring to the UCSC annotation for your genes, I don't know of any "USCD" gene predictions. And obviously if your GTF file doesn't include enhancers, you won't get counts for enhancers, so if you're interested in enhancers you'll need to include them in the GTF file.