DE comparison between different softwares

Question

[Constantine]

Hello

I'm looking for DE between a mutant and a WT cell line - 3 replicates for each condition.

I've been using 2 different R packages: edgeR and limma

If I set an FDR <0.01 and abs(logFC) > 0.3 I get for:

edgeR: ~100 DE genes

limma: ~300 DE genes

How do I assess whether the extra 200 DE genes I get in limma are actually genuine and not false positives?

When I was analsying expression microarrays I used to test whether a normalisation method was efficient by plotting QQ plots of the P.values. However, doing linear regressions on this dataset doesn't seem to be working as I expected :/

Any advice would be much appreciated

Thanks

Answer 1

[Devon Ryan]

You're presumably going to be doing some qPCR in additional samples anyway, so choose a few of the limma-only genes and see what happens. Also, have a look at whether the limma-only genes are just slightly sub-threshold in edgeR. Shifts around the border of significance are pretty common causes of things like this.

[matted]

To add to the second point, a good article on this topic is "The Difference Between "Significant" and "Not Significant" is not Itself Statistically Significant". I think it's very important to keep this fact in mind when analyzing these types of results.