A suggestion: Why use RPKM / RSEM?

[mdanb]

Thank you for this very interesting work! I noticed that you guys used RPKM and log(1 + x) RSEM values for your gene expression measurements. I don’t know about RSEM, but I know that RPKM has been shown to be biased and does not allow for intra-sample comparisons. Thus, I suggest you repeat your analysis using a more modern normalization technique like say geTMM (I’d avoid TMM alone because you want gene expression measurements to be comparable across genes, and TMM doesn’t allow for this). Just a suggestion! You might find different conclusions. Best of luck and thanks again!

EDIT: Would appreciate if you mention what normalization you used with RSEM