Question: Different results come from GO enrichment analysis with ENSEMBL ID and ENTREZID as keyTypes
0
gravatar for Jack
4 months ago by
Jack0
Jack0 wrote:

Dears,

It makes no sense for me to have different GO enrichment results just because I changed the the type of Gene IDs. Can someone help me out by explaining the reason behind this problem?

I got different results of GO enrichement analysis with different keyTypes using clusterprofiler.

I used top 50 upregulated genes with ENSEMBL ID to show this preoblem.  The organism I used was non-mode animal sheep (Ovis aries). So I created the OrgDb first. And then I used this OrgDb to do GO enrichment analysis.

library(AnnotationHub)
library(clusterProfiler)
hub<-AnnotationHub()
query(hub, c("ovis","OrgDb")) #org.Ovis_aries.eg.sqlite seems to be our orgDb
ovis<-hub[["AH66408"]] #through accession number to get the data

When I used the ENSEMBL IDs to do the GO enrichment analysis. No enrichment terms showed up.

ens.up <- c("ENSOARG00000001333", "ENSOARG00000018254", "ENSOARG00000008212", "ENSOARG00000013893","ENSOARG00000011331",
           "ENSOARG00000009412", "ENSOARG00000011128", "ENSOARG00000016625", "ENSOARG00000005493", "ENSOARG00000010832",
           "ENSOARG00000002845", "ENSOARG00000019791", "ENSOARG00000018325", "ENSOARG00000011952", "ENSOARG00000012714",
           "ENSOARG00000016983", "ENSOARG00000016920", "ENSOARG00000020162", "ENSOARG00000005835", "ENSOARG00000004205",
           "ENSOARG00000008571", "ENSOARG00000007379", "ENSOARG00000018126", "ENSOARG00000006185", "ENSOARG00000012333",
           "ENSOARG00000016101", "ENSOARG00000006969", "ENSOARG00000010787", "ENSOARG00000006789", "ENSOARG00000006474",
           "ENSOARG00000005124", "ENSOARG00000011841", "ENSOARG00000005226", "ENSOARG00000016371", "ENSOARG00000015788",
           "ENSOARG00000000540", "ENSOARG00000007053", "ENSOARG00000001849", "ENSOARG00000008891", "ENSOARG00000008859",
           "ENSOARG00000014867", "ENSOARG00000004306", "ENSOARG00000019690", "ENSOARG00000007524", "ENSOARG00000013119",
           "ENSOARG00000016353", "ENSOARG00000005981", "ENSOARG00000011280", "ENSOARG00000020521", "ENSOARG00000012641")

ego<-enrichGO(gene = ens.up,
                    OrgDb = ovis,
                    ont = "ALL",
                    keyType = "ENSEMBL",
                    pAdjustMethod = "fdr",
                    pvalueCutoff = 0.05,
                    readable = TRUE)
#no enriched terms
#no results

Then I changed the ENSEMBL ID to ENTREZID (concerns about this: 24% genes failed to map). Then I used the ENTREZID to do the enrichment analysis. Some terms are enriched.

ens.en <- bitr(ens.up, fromType = "ENSEMBL",
                  toType = c("ENTREZID"),
                  OrgDb = ovis)
#but 24% of input gene IDs are fail to map...

ego.en<-enrichGO(gene = ens.en$ENTREZID,
                 OrgDb = ovis,
                 ont = "ALL",
                 keyType = "ENTREZID",
                 pAdjustMethod = "fdr",
                 pvalueCutoff = 0.05,
                 readable = TRUE)
#got enrichment terms
#show the results first 3 terms
summary(ego.en)[1:3,1:6]

 ONTOLOGY         ID                  Description GeneRatio  BgRatio       pvalue
GO:0043604       BP GO:0043604   amide biosynthetic process     10/27 228/4322 5.241532e-07
GO:0006412       BP GO:0006412                  translation      9/27 193/4322 1.366523e-06
GO:0043043       BP GO:0043043 peptide biosynthetic process      9/27 197/4322 1.624584e-06

 

ADD COMMENTlink modified 4 months ago by Guangchuang Yu1.1k • written 4 months ago by Jack0
Answer: Different results come from GO enrichment analysis with ENSEMBL ID and ENTREZID
0
gravatar for Guangchuang Yu
4 months ago by
Guangchuang Yu1.1k
China/Guangzhou/Southern Medical University
Guangchuang Yu1.1k wrote:

Then I changed the ENSEMBL ID to ENTREZID (concerns about this: 24% genes failed to map).

It is expected to be minor difference.

You should set both qvalueCutoff and pvalueCutoff to 1 and then compare the outputs. The output IDs should be overlap at high rate and the pvalues should be correlated.

If not, report your result.

ADD COMMENTlink modified 4 months ago • written 4 months ago by Guangchuang Yu1.1k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 479 users visited in the last hour