This is a test version of Biostars. For the public version, visit https://www.biostars.org.
pvalueCutoff for enrichGO, enrichKEGG and enrichDO does not seem to function properly

When setting the pvalueCutoff (0.01) for any of the enrich functions in clusterprofiler/DOSE the resulting object sitll contains results that have a p value >0.01 all the way up to 1.00! The same is also true for qvalueCutoff.

  cnet <- enrichKEGG(cnet_list, 
  pvalueCutoff = 0.01,
  pAdjustMethod = 'BH',
  minGSSize = 10)

   cnet_desc <- cnet@result[order(cnet@result$pvalue, decreasing = TRUE), ]

Note that I have removed multiple columns in cnet_desc to show the issues

head(cnet_desc)

    GeneRatio  BgRatio     pvalue p.adjust    qvalue
    105/5239   298/8840      1        1      0.5525544
    82/5239    368/8840      1        1      0.5525544
    28/5239    453/8840      1        1      0.5525544
    15/5239    73/8840       1        1      0.5525544
    127/5239   312/8840      1        1      0.5525544
    15/5239    71/8840       1        1      0.5525544

This is seriously impacting my analysis and I'm not sure how to fix it. Any help or suggesting would be appreciated!

clusterprofiler
• 1,574 views • link
updated by Guillermo • Guillermo • written by t.mcgovern • t.mcgovern

2 answers

If I am not mistaken, all the cut offs are applied on the @result matrix to find pathways. So if you want to know which genes have been used in your KEGG enrichment, you have to filter yourself the @result matrix.

• 0 views • link
written by Bastien Hervé Bastien Hervé

I recently found this other post where they explain how to solve this issue (which, in fact, is not an error, but an incorrect way of extracting the results dataframe from the original object): https://github.com/YuLab-SMU/clusterProfiler/issues/710

• 0 views • link
written by Guillermo • Guillermo

Log in to answer this question.

More posts like this