# Which R package to use for differential analysis with TPM values?

I'm using hisat2, stringtie tools for the RNA-Seq analysis. After stringtie using ballgown I get FPKM and TPM values for every gene.

I have seen that edgeR, Deseq2 can be used for Counts data. I would like to know which R package needs to be used for differential analysis with TPM values?

Which one is better for differential analysis FPKM or TPM?

• the obvious choice would be ballgown Apr 3 '18 at 8:47
• Is using {tximport} an option? Then you could work with DESeq2. That said, it’s generally recommended not to start a differential expression analysis from TPMs, and rather use raw counts. Apr 3 '18 at 11:00
• @KonradRudolph Could you please tell me about my previous comment and why not TPM's for differential analysis? Apr 4 '18 at 9:08
• @raju I’ll refer you to Gordon Smyth’s answer. As for alternatives, you could use htseq-count or featureCounts but I strongly recommend using Salmon or Kallisto instead — in fact, you should probably consider htseq-count and featureCounts obsolete for situations where you can use Salmon/Kallisto instead (they are a lot less accurate). They will also give you TPMs but then you can definitely use {tximport}. Apr 4 '18 at 9:46