I have ChIP-seq data of RNAPII and two other factors which we think follow RNAPII during transcription in two different conditions, and I'd like to show that the genes that lose RNAPII also lose the other two factors, and same for gains.

Do you know any methods that can calculate some sort of covariation coefficients?

I know there are methods that calculate covariation of genes between two conditions, but none for different measures on the same gene.

  • $\begingroup$ do you have an example dataset? $\endgroup$
    – StupidWolf
    Nov 7, 2019 at 0:04
  • 1
    $\begingroup$ @StupidWolf Not yet, I'm still in preprocessing. I had a few issues, but I should have it soonish $\endgroup$
    – Whitehot
    Nov 7, 2019 at 9:47

1 Answer 1


Answer from @stupidwolf converted from comment:

I would try something like this, for every gene you have a logFC of RNAPII in control versus treatment. Do this for other 2 proteins, and you correlate the logFC. In very rough terms, correlation is the scaled form of covariance, so this will work.

If you have 1000 genes, do correlation between 1000 logFC from RNAPII against 1000 logFC from other proteins.

Answer from @m converted from comment:

You use a covariation stat and then perform Pearson's correlation to standardise the response between -1 to 1. An answer of >0.7 is considered robust and >0.8 highly robust. ITs an interesting question and you are right covariation is the way to answerr it


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.