Tools For Visualizing Overlap Between Go Terms

Question

[Michael Kuhn]

Some sibling GO categories are highly correlated. Is there any tool / webserver that would take a list of GO terms and show me a matrix of the fraction of shared genes in a particular species? Or at least a Venn diagramm for two or three GO terms?

For example, almost 1000 human genes are annotated as "cell cycle", but the sum of the node counts is much higher. I'd like to see which categories are overlapping with each other.

Answer 1

[Leszek]

Hi there,

Recently I have tried REViGO and I strongly recommend it. What it does is simply collapsing GO with semantic similarity. It visualise it together with nice colouring according to p-valuse (or other metric) assigned to each GO.

Some more info:

REViGO is a web server that can take long lists of Gene Ontology terms and summarize them by removing redundant GO terms. The remaining terms can be visualized in semantic similarity-based scatterplots, interactive graphs, or tag clouds.

[Pablo Pareja]

I just checked REViGO (I didn't know it existed) and I have to say it looks very promising. I should definitely take a deeper look at it when I have some time

[Thaman]

Definitely have to go through REViGO sounds promising

[Casual]

I tried REViGO with about 4k GO IDs, and made it collapsed with Http error 500 ……

[Sukhi Singh]

I think, the max limit is 500.

[Gema Sanz]

the limit is 350 terms

Answer 2

[Mary]

@Charles B. --thanks for BioVenn. @niallhaslam --thanks for the other one. I have added them to a list of Venn tools I've started here:

I ♥ Venns

There are a couple of other Venn tools over there. People here might like the Cytoscape one http://www.dishevelled.org/venn-cytoscape-plugin I haven't tried it for GO terms, but in the Cytoscape framework I would imagine it is possible.

Answer 3

[Khader Shameer]

You could try semantic similarity calculation in Ontologizer using 'Similarity' option . For a set of genes, Ontologizer generates NxN matrix with all genes in your list and can be visualized as a matrix with semantic similarity score of the two genes based on GO terms.

Answer 4

[Niallhaslam]

Hi Michael,

I've used this before for the visualisation of multiple lists of terms/ids. http://www.pangloss.com/seidel/Protocols/venn.cgi

However, if you want to go above 3 terms then you will need something more complicated. I'm currently working on something that will allow the visualisation of the overlap (with the correct spacial ratios) of 4 or more terms. If you are interested in this shoot me a mail. I'm defo interested in getting some example datasets that people need to analyse. I can send you on a poster with an example if you want.

[biotech]

This tool is not working well when using GO terms.

Answer 5

[Doron]

GOrilla is a GO-enrichment analysis tool we developed exactly to answer this kind of question. Following calculation of enrichment of GO terms, it will display the results mapped to on the GO hierarchy tree, so you can clearly see all the overlaps and relationships.

Answer 6

[Charles B.]

You can obtain a list of genes annotated by a GO terms for a selected species from [?]AMIGO[?]. It finds all genes annotated directly or implicitly (following the true path rule). If it yields more genes than it can display, you can obtain genes list in RDF-XML or tab-delimited gene association format.

When you have your lists, you can use a tool to generate a Venn diagram, like niallhaslam said. I prefer to use Biovenn which displays circles of proportional size to the sets which they represent. You can generate the figure at SVG format, which support characters drag & drop.

Answer 7

[Obi Griffith]

@niallhaslam -- Have you checked out VennMaster? This is the best tool I've found for visualizing overlap between 4 or more term lists. It is simple and works well but has a very limited set of options for customizing and often requires edits in external image editing software to produce a final publication ready imate. I would definitely be interested to see what you come up with as this is a problem not well solved so far (IMO). Venny is really good for 4 term lists (or less) but doesn't meet your criteria for "correct spacial ratios". This topic has been covered pretty well in another post.

Answer 8

[Iain]

I would recommend using the Enrichment map plugin for Cytoscape.

It can be downloaded from http://baderlab.org/Software/EnrichmentMap

Some example use cases are shown here: http://baderlab.org/Software/EnrichmentMap/Description

The plug-in generates a network view of a functional enrichment analysis, where nodes represent enriched terms and the edges between the nodes represent how similar the terms are (based on how many genes are in common between them). Interestingly, it is also possible to generate a network view comparing the results of two different enrichment analysis.