Entering edit mode
Hi,
The comparison done in the "functional enrichment for GTEx paper". I wonder if there is a way to do the comparison using enrichment results that are contained in data.frames? For example, clusterprofiler vs goseq vs topGO.
Thanks
Not quite sure why you would bother doing this because these functional enrichment tests/ packages are usually the same and based on hypergeometric distribution, the issue is just using a package that updates its definitions of KEGG or GO pathways etc. I use cluster profiler and find it very helpful.
Hi, they are actually not the same. TopGO (the R package) uses radically different ways of doing the analysis. Taking the structure of the GO-graph into account in different ways to prioritize more specific or informative results. It is quite a good improvement on the traditional hypergeometric approach. TopGO actually does do comparison of the various methods (classical vs. other approaches). But as it says below one can also do it oneself (of course). TopGO is a bit difficult to use though ...
OK thanks, that is interesting.
goseq adjusts for gene length bias, which means that the null distribution is not hypergeometric.