Normalized gene expression measurements which correlate with DESeq2 output

Question

[handgabriel]

Hi all, I've recently been analyzing some RNA-Seq data using Galaxy. I am particularly interested in differential gene expression between two groups.

I have managed to get results for differential gene expression from both Cuffdiff and DESeq2, which understandably are completely different.

My first problem is which of these results is better suited to my analysis, but I think I just need to read more about Cuffdiff and DESeq2's respective analysis methods and decide for myself, although any advice would be appreciated.

My second problem is that if I look at the output of gene expression from Cuffnorm, the pattern of gene expression matches that of Cuffdiff but not at all of DESeq2 (e.g. the gene with the highest log fold change in DESeq2's output was not detected in any sample in Cuffnorm's output).

Is there a way I can generate normalized expression levels for each sample and each gene using the same method that DESeq2 uses rather than using Cuffnorm?

Thank you in advance.

Gabriel

Answer 1

[Devon Ryan]

In the Galaxy wrapper for DESeq2 there's an option to output normalized counts ("Output normalized counts table"), just set it to "Yes".

I would always trust the results of DESeq2 over those of cuffdiff. There's a reason that DESeq2 is one of the standard analysis programs and cuffdiff is not.

[handgabriel]

Thanks Devon that's helpful, I was along the same lines of thinking about DESeq2, but I'm no bioinformatician :P

Anyway thanks again for the advice.