Recommended approach: gene set analysis of small transcriptome experiments?

Question

[Nick]

We have RNA-seq data for a small experiment, which compares the transcriptome of a treated vs untreated cell line (3 biological replicates for each condition, 6 samples in total).

So far we have obtained lists of differentially expressed genes using DESeq. We want to perform gene set analysis to identify pathways which are dysregulated (enriched for differentially expressed genes). However, I have read (e.g. here) that many established gene set enrichment methods work by permuting sample labels to generate a null distribution.

Given our very small number of samples, this is presumably not an option. Are there gene set analysis methods specifically able to cope with such small sample sizes?

Answer 1

[Benn]

You can try goseq, it is meant for RNA-seq data and uses length-bias correction with a Wallenius distribution (as default).