Getting interesting gene ontology analysis result

Question

[bharata1803]

I want to ask about gene ontology analysis.

So, I have around 1,200 transcription factor genes ID (EnsEMBL id). I have done some clustering on a different constraint and I want to see the result of my cluster in the GO context. The only problem here is because all of them transcription factor, I always get transcription factor related ontology as the top. Because all of them are TF, of course ontology analysis will produce the result. What I want is to find ontology other than that so I know what kind of function this cluster works together, for example related to cancer development, stem cell, etc.

My question is, what kind of tools or script that I can use or make to filter out the transcription factor related ontology? Now, I am using DAVID and several other online tools. If possible, I want to develop my own script in python so that I can easily repeat that GO analysis for all my cluster that I found.

[venu]

I would go with suggestions from @Sean. One useful resource would be tftargets

[theobroma22]

Since you have the EMBL IDs you can convert them to Entrez IDs, the use the Bioconductor package SPIA, although you will use R.

Answer 1

[Sean Davis]

I suspect that the answer to your question is that you need to map your transcription factors to their putative targets and then perform your gene ontology analysis based on the associated target genes as mapped to your clusters.

[bharata1803]

So, what I've done is to create transcription factor network. What I do is I find target gene of a transcription factor that is also transcription factor. From this, I have a network of Transcription factors. You can imagine a chain of transcription factor that regulates each other. TF A regulates TF B, TF B regulates TF C, and so on create a large and complex network.

I want to see what is the most interesting ontology that I can extract if I select a cluster of nodes that are connected, or we can say a sub-network. The question would be finding the most interesting sub-network in the gene ontology term context.

I have downloaded GO annotation file (http://geneontology.org/gene-associations/goa_human.gaf.gz) and GO term database (http://geneontology.org/ontology/go-basic.obo).

So, I think I just need to exclude transcription factor related term to be inputted into the GO analysis tools.

What do you think about that?

[theobroma22]

I personally don’t think the GO results will be interesting, or useful for that matter. Hope this helps.

[bharata1803]

I also use cancer log fold change in the network to filter the nodes. My network now is a differentially expressed transcription factor network.

Given nodes that are connected (subnetwork), I want to see what are the ontologies that are involved in that subnetwork and because all the nodes are differentially expressed, it probably has biological meaning.

I have check some nodes in the network and I get some interesting ontology, for example: negative/positive regulation of cell proliferation, cell differentiation, multicellular organism development, etc.

Cell proliferation mean that the up/down regulation of that TF probably has some role in the cancer development.

The aim is to understand what are the probable effect of up/down regulation of transcription factor. The network can represent the chain effect that is happening or maybe causal relationship.