Different results between clusterProfiler and Enrichr

Question

[abbas89]

Hi, I first used enrichr and found an intesresting pathway enriched in my geneset. Then I wanted to use clusterProfiler with the same geneset but this time including background genes. My issue is that clusterProfiler gives completely different results compared to enrichr even when not including backrground genes in clusterProfiler. Of course I used the same database in both cases (bioplanet).

em <- clusterProfiler::enricher(genes, TERM2GENE= bioplanet,pAdjustMethod="BH", universe=background)

Is it expected to have so much discrepancy between these two widely used tools?

[Jean-Karim Heriche]

but this time including background genes

So you're doing something different with a different tool. Why would you expect the outcome to be the same? If you're concerned about this then you should understand what each tool is doing, from the data they use to how it's processed and which algorithm they implement.

[abbas89]

That's why I said "even when not including backrground genes" in clusterProfiler I still get completely different restults.

Answer 1

[ATpoint]

Yeah, well, welcome to the mess that we can "enrichment analysis". I cannot answer the question directly, but would generally recommend against EnrichR as it is a blackbox that cannot be run locally and hence strictly speaking is not reproducible. I use clusterProfiler as it runs locally, has reproducibility and accepts a background which is critical imo.