Entrez Gene ID

Question

[synat.keam]

Dear Seniors and all members,

Me again!! I hope you do not mind me as a junior in RNAseq and tried to learn and finish my degree. Sorry for another question. I have done WGCNA and was able to identify the module associated with traits and exported data for Gene Ontology and network via cytoscape. I worked on mouse tumour tissue untreated or treated with radiation.

I had a module, which contains almost a 1000 genes and I would like to do gene ontology particularly over representative analysis for that module. Basically, I had gene symbol in my output and to do gene ontology, I had to convert it to entrezID. I use the following code chunk from Biomart library.

gene_id<- getBM(attributes = c("ensembl_gene_id", "external_gene_name", "entrezgene_id"), mart = useDataset("mmusculus_gene_ensembl", useMart("ensembl")))

Here is the output I got.

and I used the following code chunk to map entrezID to my gene symbol

d1<- merge(d, gene_id[,c(2,3)], by.x="GeneID", by.y="external_gene_name")

## Gene Ontology

I first used ClusterProfiler to perform GO over representative analysis using the following code after creating genelist.

ego <- enrichGO(gene = gene, universe = names(geneList), OrgDb = org.Mm.eg.db, ont = "BP", pAdjustMethod = "BH", pvalueCutoff = 0.01, qvalueCutoff = 0.05, readable = TRUE)

I got error as shown in the following

--> No gene can be mapped.... --> Expected input gene ID: 75002,70099,50796,104310,71099,18663 --> return NULL...

I guess I had the correct entrezID in my genelist, which is integer. Why there was no gene that can be mapped? I had read previous posts in biostar from members, who had experienced the same problem and our members said these are not mouse entrezID. May I ask whether my code chunk used to generate generate entrezID is correct for mouse organism.

Also, I have try another option in R using

library(GO.db)

library(GOstats)

selectgene<- unique(d1naf['entrezgene_id'])

unversegene<- unique(d1naf$entrezgene_id)

cutoff= 0.05

testing<- new("GOHyperGParams", geneIds= selectgene, universeGeneIds= unversegene, annotation= "org.Mm.eg.db", ontology= "BP", pvalueCutoff= cutoff, conditional= FALSE, testDirection="over")

GOtest<- hyperGTest(testing)

summary(GOtest)

Here is the output

summary(GOtest)

1 GOBPID Pvalue OddsRatio ExpCount Count Size Term

<0 rows> (or 0-length row.names)

Warning message: No results met the specified criteria. Returning 0-row data.frame >

Does this mean my genelist is not found in any GO geneset or just because the error with entrezID conversion? it is weird because I had lots of genes and I should have gotten some outputs. Thank for taking time reading my question. Is there any online web-interface to do gene ontology outside R? looking forward to hearing from all seniors and members soon.

Kind Regards,

Synat,

[GenoMax]

Looks like you hit Save before completing the post. You can edit and complete.

[synat.keam]

Dear GenoMax,

Thank you so much for your quick response. I had error and now completed my post. wondering if you could have a look and suggest sth.

Kind Regards, synat

Answer 1

[R_newbie]

I noticed you have a lot of NA in your id, try to remove them before loading to the profiler, see if it helps

[synat.keam]

HI R_newbie,

thanks. I have removed all missing value for using clusterprofiler. not working at all.

thanks

[R_newbie]

since I don't have your sample data, I can't really test the code. But I noticed that your selectgene are unversegene are the same? could that be a problem.

Universe means all genes expressed in your sample,

select gene is the ones which you select such as differential gene?

selectgene<- unique(d1naf['entrezgene_id'])

unversegene<- unique(d1naf$entrezgene_id)

[synat.keam]

Hi R_newbie,

Thank for your email. Basically, this is not differentially expressed gene from DEseq2 or limma at all. This is the gene module from WGCNA (kindly view the picture I attached)

In this case, let say I am interested in black module. Therefore, i would like to examine whether genes in black module are over-represented in GO term. So I am assuming universal gene are genes of all combined modules? What do you normally do in that case? this is my first time doing go with gene module from WGCNA. also, in this case, I do not have logfold change, but MM value instead. However, I think it does not matter with the value because this is over-representation test and as long as we have genelist then shoudl be okay?

I also I would like to insert another question whether we can just do gene ontology via Gene Ontology web-based (kindly review attached) and link: http://geneontology.org/. Not sure I did it correctly, but I just pasted the genelist from the black module in the box and clicked launch and then it directed me to the result page. However, I could not find anywhere asking me to provide universal gene. What do you think aobut this way? right or wrong? you have any suggestion how I could do this using WGCNA module. Looking forward to hearing from you soon.

Regards,

synat