RNAseq data analysis: FPKM and DESeq2

Question

[MV]

Hi to everybody, Maybe I have a silly question but I'm new in this field. An external company has sequenced my RNA samples that corresponds to two different cell lines and they give me back the results data relative to gene expression. For the first cell line they calculate the fold change from FPKM data whereas in the second cell line they use DESeq2. I think that the only way to compare them is starting again the analysis with DESeq2 from the counts ora maybe there is a function in DESeq2 that work with FPKM data.. I don't know how to menage because I don't have the row data.. There is a way to compare those results?

Thank you in advance!

[MV]

Thank very much Kevin, very useful!!

Answer 1

[Kevin Blighe]

That's a fairly 'shoddy' way to do an analysis on the part of this company, no? I would send it back to them and ask them to re-do everything using DESeq2 and to return the DESeq2-normalised counts and both the variance-stabilised and regularised log expression levels. They could also produce FPKM expression levels across all samples and genes. Be sure to request all code that they use.

DESeq2 cannot do anything with FPKM data - it needs raw counts.

The only way that you could work with the fold changes calculated from FPKM expression levels and DESeq2-normalised counts is via a correlation analysis.

Additionally, if you have the FPKM expression levels count matrix, and also the DESeq2-normalised count matrix, you could possibly aim to standardise these to Z-scores, but this is a bit messy and would not really stand in a published manuscript (unless the bioinformatics methods were hidden from reviewers, as is the case in most published works).

Kevin