Question: Is logFC in edgeR base e, 2, or 10?
0
gravatar for O.rka
6 months ago by
O.rka50
O.rka50 wrote:

I am trying to find in the documentation of edgeR where it says the default base of the log function. I've used exactTest and topTags it gives me logFC but is it base e, 2, or 10?

ADD COMMENTlink modified 6 months ago by Devon Ryan84k • written 6 months ago by O.rka50

DE analysis by edgeR

I read here that is likely base e 2

ADD REPLYlink written 6 months ago by F3.1k

DE analysis by edgeR

I read here that is likely base e 2

ADD REPLYlink written 6 months ago by F3.1k

It's definitely 2 in the topTags

ADD REPLYlink written 6 months ago by russhh3.8k
4
gravatar for Devon Ryan
6 months ago by
Devon Ryan84k
Freiburg, Germany
Devon Ryan84k wrote:

It's log2 in all common RNAseq differential expression programs, including edgeR. The reason for that is that counts can range from 0 to infinity, so using a log2 transform allows running the MLE optimization in (-infinity, infinity), which is simpler and better for fold-changes <1.

ADD COMMENTlink written 6 months ago by Devon Ryan84k

I'm not sure this is an explanation for why log2 is taken instead of log10 or the natural log. Indeed, I'm fairly sure that the coefficients that are stored in the fits in edgeR use the natural log, consistent with most GLM approaches; it's only when converting to human readable form that the base is shifted from e to 2

ADD REPLYlink written 6 months ago by russhh3.8k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 626 users visited in the last hour