gene expression comparison between tpm and fpkm

Question

[jabbari.parnian]

Hello everyone,

I have FPKM and log2 transformed TPM data of two different experiments. I want to compare the gene expression between these two experiments (I know it is not optimal but I don't have access to the raw data to calculate gene expression of both experiments in FPKM or TPM format). I have read the previous threads and they suggested that log2 transformation ca cover for this bias to some extent. I was wondering if that is right? can I compare the differentialy expressed genes of these two experiments with one another?

Thanks,

Answer 1

[ATpoint]

I think not. These are different experiments, normalization methods are different and data are therefore probably confounded. Be careful.

[jabbari.parnian]

even if I log2 transform the fpkm data too?

[N15]

You could compare genes within one dataset, i.e., Gene A across samples within the TPM dataset and Gene A across samples from the FPKM data set, but I would NOT directly compare normalized values between the FPKM and TPM data sets. Get the raw count data and ensure they are normalized in the same manner.

Also, I would transform both data sets in the same way. So if you log2 transform TPM, I would also log2 transform the FKPM set.