Entering edit mode
3.4 years ago
Rob
▴
170
Hi friends I am doing gene set enrichment analysis with GSEA java base software from broad institute. I have 44 patients (2 groups: 22 control vs 22 treatment) and 19000 genes. How should I interpret the result? No gene has FDR less than 0.05. Why is this? there should be some significant genes but there is not.
Why do you expect an effect? Do you have independent evidence for it? Is it known that the treatment affects gene expression?
I assume you mean that none of the FDR-adjusted p-values are below 0.05. All that this result says is that by accepting no more than 5% false positives, you don't find any enriched gene. That's what you can conclude from this experiment. However, there's nothing special in using the value 0.05. You should set the value based on circumstances and the cost associated with making a mistake. I've happily used 0.1 when the purpose was to find candidates for follow-up experiments. Another approach is simply to rank the genes by magnitude of the effect and pick the top x for validation. For example choose x based on how many genes your resources allow you to test in independent experiments. Occasionally one may be in a "pathological" case of p-value distribution. This can be detected by plotting a histogram of the uncorrected p-values. See here for more details.