The quest for the perfect gene-set analysis software

Question

[roy.granit]

I've been dealing with gene-expression\gene-set analysis for about 6 years now, started with micro-arrays and recently RNAseq data, yet I still haven't found a convenient & reliable software to handle these tasks. I turn to you BioStars -perhaps you can point me at some neat tools I have been missing all these years? :)

The basic task I wish to conduct is to assess the statistical significance of changes to various custom gene-sets I input, clustering & graphical representation of the output is a nice bonus, but not a must.

These are some the tools I'm currently using:

• Genomica - Good old Genomica is piece of software, and does the job very nicely with hyper-geometric distribution! the only trouble is that's it's really outdated, and has lot's memory issues which makes it really hard to load large amount of data. Also, one cannot just copy-paste gene sets, but have to go thought some laborious work to format the sets.
• GSEA - On first sight it looks like a winner, but I find it almost impossible to get it to run properly, and I simply hate formatting the input data (and the whole java interface).
• Expander - Seems promising, but not that intuitive to use (I was not able to download the most recent version).

Thanks for taking the time to read through, open to all suggestions.

Roy

edited:

P.S - please refer me to non-commercial software

[Sean Davis]

Sounds like you have additional constraints? GUI? Cross-platform? What databases included?

[roy.granit]

I'm using a Mac, but have access to a PC if needed. A nice GUI is a plus of course, but if you know a great package that can make this easy that's also OK. I'll be uploading my own gene-sets, so no specific DB is required (most of my sets are already in Entrez ID, so it would be nice to keep that).

Answer 1

[Woa]

http://www.advaitabio.com/ipathwayguide.html#page=page-1

[roy.granit]

Thanks, but I was thinking of non-commercial ones.

Answer 2

[Woa]

This is a good review:

"Methods and approaches in the topology-based analysis of biological pathways"

Answer 3

[Jean-Karim Heriche]

Why not a DIY solution in your preferred programming language?

[roy.granit]

I think that since so many people are doing this it would be better to build upon some well tested robust code.

Answer 4

[Zhilong Jia]

R package, cogena, probably is one of your options. The custom gene-sets is accepted if formatted as gmt (so far, only gene symbol is supported due to the internal gene sets. but if your input genes and gene sets are all Entrez IDs, cogena should work.) Graphical representation of the output is also a feature of cogena. I do not know what's the meaning of clustering here, but cogena cluster DE genes and the do pathway analysis for the clustered genes as well as All the DE genes.

See post Drug repositioning based on co-expressed gene set enrichment analysis, cogena also.

Another tool is enrichr, though the custom gene set is not allowed so far.

Hope helpful.

[roy.granit]

Thanks! as for cogea, looks cool - if I understand correctly it requires one to create some clusters and then preform the gene-set analysis, right? or can you skip the first step?

[Zhilong Jia]

Cogena ​clusters genes internally. If you just want to do pathway analysis of all de genes, cogena shows the result always , together with pathway for clustered genes.

Answer 5

[EagleEye]

How about GeneSCF,

1. More reliable because it holds up-to-date information (also real-time features). Specifically designed to avoid using outdated databases/information.

2. User friendly (computation biologists)

3. No installation

4. Can be easily integrated with any pipeline

5. Support for all organisms covered by KEGG and gene ontology

Answer 6

[roy.granit]

Now that it's been >2.5 years maybe some has some cool software to recommend?