Gene Set Enrichment Test Tool That Can Compare Multiple Gene Lists

Question

[confusedious]

I received some excellent recommendations of enrichment test tools for testing single gene sets here at Biostars in the past. I am hoping someone will have something similar to offer for my latest query.

I have three gene lists (~1600 genes) that are not ranked in any particular way and have no expression data (they were obtained based on the Fst of SNPs within them - I have dealt with linkage disequilibrium to the best extent I am able, so not too worried about this). I would like to compare these three gene lists for enrichment of biological processes, in particular I am interested to see if they are enriched for immunological function.

I have become familiar with DAVID, but the issue I am finding is that I have to run the three tests separately and the enrichment scores obtained for gene functional classification seem to be only internally meaningful (that is to say comparing one gene list to another isn't a simple matter of side by side comparison in terms of enrichment scores).

Does anyone know of a test that can run all three lists side by side and statistically compare them?

[Bharat Iyengar]

there is no tool that I know of. But if you are familiar with any programming language, then it is an easy task.

[confusedious]

Truly? I have a novice level grasp of R - are you aware of any way this could be done simply here? I'd be very interested to hear, as this step of my research has been a bit of a bottleneck on my workflow.

[confusedious]

Actually, as PANTHER is calling over representation from set gene lists, that may work well for my purposes. I can compare these tests side by side by p-value - I think this may do the trick.

[azimzadeh.jam.sadegh]

Dear confusedious;

I need the same thing but for several hundred tests. Can you share your current experiences? Regards

[Istvan Albert]

You should ask this as a separate, new question, otherwise it will be buried deep into another question where very few people will see it.

Answer 1

[David Westergaard]

If you are familiar with R, you could use the Bioconductor package ReactomePA (Reactome Pathway Analysis). Alternatively, you could also just use the Reactome web interface and choose an overrepresentation analysis. Both use a hypergeometric statistic to test for statistical overrepresentation. If by "Biological Processes" you are refering to the Gene Ontology category, there is also the package topGO, which offers a few other statistical tests rather than just the Hypergeometric test. For all of these methods it should be possible to save the output into a tabular format for easy comparison. If you are familiar with ggplot2, you could also produce a dot chart where dots size is proportional to -p value, to make it even easier to compare.

[confusedious]

This is straight up and down fantastic. Thank you for the recommendation. The web interface is excellent and the output it produces is straight forward to read and also lumps together the immune system in a way that works for me. I'll give the R package a try over the weekend.

Answer 2

[Woa]

Ingenuity pathway analysis and (probably Metacore GeneGO too) has the option of loading multiple gene lists for pathway enrichment analysis. They are not free, however you can download trial versions that lasts from one/two weeks.

Though you're not using gene expression data as such, but have a look at this paper if it comes handy somehow:

• http://nar.oxfordjournals.org/content/38/17/e169.abstract
• http://systems.crump.ucla.edu/rankrank/

[confusedious]

I have access to Metacore in my lab - I haven't tried it as of yet so I was unaware of this property. Thank you for the input, I'll take a look.

Answer 3

[azimzadeh.jam.sadegh]

Dear confusedious;

I need the same thing but for several hundred tests. Can you share your current experiences? Regards