Why enrichGO does not return any enrichment terms?

Question

[Firingam]

Hi! I'am having some issues with clusterProfiler::enirchGO. I provide down below a partial list of my genes in order for you to have the possibility to do a reprex

Genes <- c("8863" , "5588" , "9612", "83786","57661", "4093" , "6141" ,"253943","4628" ,"10495", "3688", "6399")

They are ENTREZID Now by applying enrichGO:

GOtest <- clusterProfiler::enrichGO(gene= Genes,

                      OrgDb = org.Hs.eg.db,

                      ont = "MF",

                      pAdjustMethod = "BH",

                      pvalueCutoff  = 0.01,

                      qvalueCutoff  = 0.05,

                      readable      = TRUE)
  # They are Human ENTREZID

As far as the output is concerned:

# over-representation test
#
#...@organism    Homo sapiens 
#...@ontology    MF 
#...@keytype     ENTREZID 
#...@gene    chr [1:535] "8863" "5588" "9612" "83786" "57661" "4093" "6141" "253943" "4628" "10495" "3688" "6399" 
# "60436" "7442" "203259" "55588" "339983" ...
#...pvalues adjusted by 'BH' with cutoff <0.01 
#...0 enriched terms found
#...Citation
   Guangchuang Yu, Li-Gen Wang, Yanyan Han and Qing-Yu He.

clusterProfiler::groupGO, even if is different, performs quite well. So, why I'm not having any GO term returned to me?

Thank you in advance

P.S. : obviously the gene list I'm working with is far more larger than the one I provided

[i.sudbery]

Perhaps no gene sets are enriched?

[Firingam]

So do you think it is independent from the script?

[Kevin Blighe]

To test, please adjust your p- and q-value thresholds to 1.

[i.sudbery]

The script looks fine to me. Its not impossible something else is wrong, but a genuine lack of enrichment is by far the most likely explanation.

[Papyrus]

do both functions adjust the p-values for multiple-testing and filter by those?

[Firingam]

Do you intend to change BH to another kind of test?

[Papyrus]

No, I meant to say the same thing that Kevin (much more clearly) stated: that you should set the same p-value, q-value thresholds (to 1, to capture all the results), to really compare, because the functions may be filtering the results differently.

If, after checking that the functions are being used in a similar manner, you have many significant results in one of them, if the two functions perform comparable analyses (check this), one could expect to have comparable results. However, if you have borderline results in one of them, it could be normal to not detect enrichments in the other.