ChiP Seq differential analysis with Homer

Question

[B612]

I'm using Homer to perform a differential analysis between KO and WT with ChiP Seq data.

I follow this protocol:

1. makeTagDirectory with the different bam files
2. Run annotatepeaks.pl in this order: KO KO WT WT
3. Run getDiffExpression.pl tablecounts-txt KO KO WT WT (following the same order than I have in my table obtained in the step before)

At the end I want to plot the Fold Change of KO over WT (ratio KO/WT) but I don't fully understand how this is calculated by HOMER with getDiffExpression because the output is :

Output Stats KO vs. WT:

Total Genes: X

Total Up-regulated in WT vs. KO: X ( X %) [log2fold>1, FDR<0.05]

Total Dn-regulated in WT vs. KO: X ( X %) [log2fold<-1, FDR<0.05]

This make me feel confuse about the sense/direction of the ratio (Fold change). Is it doing the KO / WT or WT / KO?

I think understand this is very important to be able to interpret correctly my plots and I don't find the documentation of Homer clear in this regard. Also, I took a look to different papers that use this methods, but didn't explain in methods how they exactly apply getDiffExpression.pl, if they used the order mutant mutant wildtype wildtype or the other way around.

Can someone help me?

Thank you very much!

Answer 1

[biomon]

Hi B612, I just had a look at the manual and I agree, it's not clear. From your output I am assuming WT vs. KO was performed. I would check the the coverage or bigwig files for top differentially expressed promoters (??since you are looking at number of genes), and confirm that it is WT vs KO.

Also it is confusing that it says KO vs WT at the top.

One thing I did notice in the manual is that the control conditions are put in first and then the treated (annotatePeaks.pl ...etc.. ctr/ctr/treat/treat >counts.txt and getDiffExpression.pl ....etc... ctr ctr treat treat). So I would assume in the most basic form of ctr and treatment, the output should give you Treatment vs Control in logFC. And in your case you put KO then WT, so following this logic it should give you WT vs KO. Again not very clear, I am only assuming, so the best thing to do is check the bigwig.