If changing the RNA-Seq count normalisation method changes the DGE and downstream analysis results?

Question

[analyst]

Hi,

My RNA-Seq count data is FPKM normalised. But reviewer is asking to convert FPKM data to TPM normalised one.

Now If I perform TPM normalisation, and use this TPM normalised count file to perform DGE analysis will it change the analysis results? My question is that after converting FPKM normalised data to TPM normalised, do I need to repeat whole DGE analysis and downstream analysis steps again or not required?

Thanks alot!

[i.sudbery]

See my detailed answer to this here: Why employ normalization methods, and how can they be utilized in DEG analysis?

[analyst]

Thank you i.subery!

[dsull]

How are you performing the DGE analysis / what tool are you using for it?

[analyst]

Thanks dsull, I am using deseq2 for DGE analysis

Answer 1

[dsull]

OK, since you mention you're using DESeq2 for DGE analysis.

DESeq2 uses counts (not FPKMs, not TPMs) for DGE. So if you're using DEseq2 correctly (i.e. NOT giving it FPKMs), then you're fine -- there's nothing more for you to do.

If you ran DESeq2 with your FPKM values, then you need to repeat your entire DGE analysis because you did it incorrectly. DESeq2 was not designed for FPKM values (it is not designed for TPM values either; you need to give it unnormalized "counts").

[analyst]

Okay thank you so much for making things clear. Raw counts were used for DGE analysis in DESeq2.

Thanks a lot for saving time dsull!