search

ctlab / fgsea

Fast Gene Set Enrichment Analysis
Other
369 stars 65 forks source link

fgsea hangs with nperm=1001, but not 1000 #93

Open assaron opened 3 years ago

assaron commented 3 years ago 
# Runs fine, takes just a few seconds
fgseaRes <- fgsea(examplePathways, exampleRanks, nperm=1000, maxSize=500)

# Hangs and never completes
fgseaRes <- fgsea(examplePathways, exampleRanks, nperm=1001, maxSize=500)

It likely indicates that there are problems with BiocParallel package and it has to be reinstalled. In this case the following code hangs as well:

library(BiocParallel)
example("bplapply")
assaron commented 3 years ago

@ChristianRohde please check if the BiocParallel example above works for you or not.

ChristianRohde commented 3 years ago

I cannot reproduce the unexpected behavior neither on my installations on Ubuntu, MacOS or Windows any more. No need to add the nproc=1 any more as well. I could swear that I got stuck on this way too long yesterday, but now it is gone without a trace. I do not remember if I updated something on all 3 computers, but at least fgsea is now just working again as before very fast. Thank you for your support!

mvhunter1 commented 3 years ago

Hi, I'm also having the same issue with fgsea running infinitely, seemingly out of nowhere since I had used the function successfully for at least a year before this. Setting nproc = 1 doesn't seem to help. The BiocParallel example above does work for me. I'm using the latest version of the fgsea package on MacOS. I (used to) rely heavily on this package so any advice would be really appreciated!

assaron commented 3 years ago

@mvhunter1 Is it the problem of some particular dataset or it hangs on different inputs?

mvhunter1 commented 3 years ago

Hi, it hangs on any input I try from multiple datasets.

Edit: this seems to be specific to fgseaMultilevel as fgseaSimple seems to work fine.

assaron commented 3 years ago

@mvhunter1 Does it fail for the example dataset? Does setting nperm=1000 help? It looks like a different problem, not the one described in this issue, so please create another issue and describe exactly what works and what doesn't work for you providing the datasets.