Search
Question: clusterProfiler : How works the q-value filtering ? What is a good cutoff ? Is this like an FDR ?
0
2.1 years ago by
ZheFrench10
ZheFrench10 wrote:

Im  using clusterProfile clusterProfiler_3.0.5 on R  3.3.1  as follows :

kegg <- enrichKEGG(entrez_id, organism="hsa", pvalueCutoff=0.05, pAdjustMethod="BH", qvalueCutoff=0.2,use_internal_data=FALSE)

write.csv(summary(kegg),file=paste0(c(getwd(),dir_pathway,"DESEQ_KEGG_ENRICHMENT.csv"),collapse="/"))

I have two questions:

I don't understand how works the pvalue and qvalue. pvalueCutof is just use for filtering on pAdjusted value or raw p-value ?

In my output, I have  lines with a qvalue upper to 0.2 whereas my qvalueCutoff is set to 0.2. Is this a bug or qvalueCutoff is used for another thing than simply filtering here.

What is q-value ? I set to 0.2 (taken from an example) But is it a kind of FDR...?  Kind of relative to that conceptual question about FDR, FDR adjusted p-value and q-value

modified 2.1 years ago by Guangchuang Yu1.1k • written 2.1 years ago by ZheFrench10
1
2.1 years ago by
Guangchuang Yu1.1k
China/Guangzhou/Southern Medical University
Guangchuang Yu1.1k wrote:

If you set qvalueCutoff = 0.2, it will filter out those larger than 0.2. https://github.com/GuangchuangYu/DOSE/blob/master/R/enricher_internal.R#L147