Does BioMart based go-terms contain any insignificant go-terms ?

Question

[arnetmit]

Hi ,

I want to obtain significant GO-terms of gene lists that I have already had. I used BioMart R package and it returns all GO-terms of gene lists but it does not supply any p-value (fisher exact test for instance ) or fdr value to eliminate insignificant GO-terms. I researched on the internet but I haven't found any information about how BioMart presents GO- terms without any criteria. Does BioMart based go-terms contains any insignificant GO-terms when we used our gene lists as parameter ? Does BioMart return GO- terms (without any test) whether they involve in any GO- terms or not ?

Thank you for your interest.

Answer 1

[Jean-Karim Heriche]

BioMart doesn't perform any analysis. It simply gives access to the annotations, i.e when you query BioMart with a list of genes to get GO terms, it returns all the GO terms the query genes have been annotated with.

[arnetmit]

thank you for the reply. Does BioMart based go-terms are significant ? or they are only GO-terms in which genes in our gene list involve in ? The last question, GO-terms of all genes in the gene lists retrieved using genes separately or all together ?

[BioinfGuru]

"Does BioMart based go-terms are significant ?" No - significance is not considered at all

"Are [they] only GO-terms in which genes in our gene list involve in?" Yes - you tell biomart a gene name, biomart tells you the go-terms that are associated with that gene name (there is no test for significance)

"separately or all together?"- Seperately

Biomart is just a tool for getting information from a database of annotated genes. It does not consider the relationship between the genes in the list. It does not test for significance. It sounds like you are looking for a tool like g-profiler which does do those things.

[arnetmit]

Finally I found exact answer. Thank you so much.

[WouterDeCoster]

It's pretty much the same as Jean-Karim said.

[arnetmit]

You are right and I had marked Jean-Karim's answer as solution before you stated it. Thank you all. ( @Jean-Karim and @kennethcondon2007)

Answer 2

[theobroma22]

There is a P-value associated with the gene enrichment of each determined GO term. So, there is an expected number of genes for each GO term, and the enrichment can be more or less than what's expected. As for any other analysis, go ahead and choose a P-value cut-off like 0.05 to demonstrate significance.

[WouterDeCoster]

Biomart doesn't perform enrichment analysis on GO terms. Your answer is not correct. See answer of Jean-Karim for explanation.

[theobroma22]

"Please be mindful of language barriers of the OPs." I know biomart very well. This is why my answer is about gene enrichment specifically and NOT about BiomaRt. But, yes I see your point.