gseGO: no term enriched under specific pvalueCutoff

Hi all,

I have attempted to run gseGO as follows:

gseaResults <- gseGO(geneList     = rankedList,
          OrgDb        = org.Mm.eg.db,
          ont          = "BP",
          nPerm        = 1000,
          minGSSize    = 10,
          maxGSSize    = 500,
          pAdjustMethod = "BH",
          pvalueCutoff = 0.05,
          verbose      = T,
          by = "fgsea")

Where, "rankedList" is a sorted double with logFCs named after EntrezIDs like so:

   77583   234564   228802    11418    79235    21414 
9.421894 9.089247 7.089965 6.471895 6.298729 6.141589

This includes 11234 genes in the results of DESeq2 after gene name conversion via bitr.

However, I could not get any result.

no term enriched under specific pvalueCutoff...

I have also tried it with ENSEMBL instead of ENTREZ. My clusterProfiler version is 3.14.3 and R is version 3.6.2.

Thank you

clusterprofiler

• 12,369 views • link

updated 3.1 years ago by Ram

• written 6.3 years ago by 14lwt1 •

Hi 14lwt1,

I have the same issue. If I set the pvalueCutoff = 1, then I get the full lists from which I can see that several biological processes do actually have pvalue<0.05. Hence, the argument pvalueCutoff seems to me to be non-functional.

• 0 views • link

written 6.1 years ago by rikki.frederiksen •

pvalueCutoff is use to filter pvalue and p.adjust.

• 0 views • link

written 5.0 years ago by Guangchuang Yu

1 answer

Your code is fine. You can try to set pvalueCutoff = 1 to get all possible results. You need to check your data at DESeq2 stage, for example, MA plot and vocanol plot, PCA or hclust to check sample distance.

• 0 views • link

written 6.3 years ago by MatthewP

Thank you for the reply. Yea, I was just surprised that there isn't a significant result with that many genes as input. There are differences across the comparison based on sample distance.

• 0 views • link

written 6.3 years ago by 14lwt1 •