Entering edit mode
6.8 years ago
wayj86
▴
40
Hi all,
I am using STAR to align my reads to human genome. Then I need to calculate the TPM values of genes. There are many software that can be used to do that: Stringtie, RSEM, kallisto, Salmon, salfish, ...
So which one is best for doing that?
Many thanks, Stanley
Do you definitively need to just calculate TPM values or do you ultimately want to perform a differential expression analysis?
If you have aligned your reads to the genome, then you should have SAM or BAM files.
You can calculate raw counts over these SAM / BAM files using featureCounts, and then read the raw counts into EdgeR or DESeq2 where you can perform normalisation.
Thank you very much for your answer. I need TPM to perform WGCNA, actually.
WGCNA can take any type of normalised and/or logged data, in fact. Were you told to just use TPM? WGCNA is fundamentally based on correlation.
Yeh. WGCNA usually takes FPKM data, but I was told TPM was much better than FPKM. That is why I need to calculate TPM.
I do not believe that's true at all. WGCNA takes any data:
One could argue that TPM is better than FPKM (for differential expression analysis), but neither of these are better than TMM or the geometric median/mean methods used by EdgeR and DESeq2, respectively.
In any case, if you use featureCounts to derive raw counts, like I mentioned, you can then convert these to TPM: Calculating TPM from featureCounts output