Entering edit mode
2.8 years ago
Stevens
▴
20
Hi, I have KO and WT cells that have been pre-analyzed and I am attempting to use the R package iDEA to identify biological pathways involved in the behaviour of my KO cells.
iDEA requires a gene specific annotation file that has matching gene names to the DE file created during pre-analysis, however I am finding it difficult to find a gene specific annotation file that works with my data or that is specific to mice
I have found a couple of resources but I am not sure how to utilize them either.
- http://bioinf.wehi.edu.au/MSigDB/
- http://bioconductor.riken.jp/packages/3.8/data/experiment/vignettes/gskb/inst/doc/gskb.pdf
Any information about possible gene specific annotation files is appreciated.
Looking at the iDEA manual, they do provide a link to download pre-formatted mouse gene set annotation files (at least for GOs).
I see that those files contain genes with the MGI ID nomenclature. If your DE file contains genes with a different nomenclature, you could use the R package biomaRt to convert them to MGI IDs.
If you want a database with many gene set files from the MsigDB (GO, KEGG, Reactome...), you could try the R package msigdbr. The gene sets are annotated with entrezgen ID, ensembl ID and gene symbol. But you would have to format its files into the format used by iDEA.
Thanks, I have tried out the dataset within R, its names aren't overly informative so I'm also testing out a dataset from msigdb. The result from idea's mouse gene set didn't return any pathways at a significant level so I'm trying to work out if something's wrong with my data or the gene set now. Alot of my p-values in the DE were 0 from being too low so that may be the problem.
OK, there is always the case that there are in fact no enriched pathways. Aside from the programmatic approach, you can get a quick overlook of which pathways you would expect to be enriched if you submit your list of significant genes to the Enrichr web (it looks across many many databases).
I've considered that, but there were alot of differentially expressed genes so even by chance theres likely to be a few pathways. I'm not entirely sure how to navigate enrichr but it gave alot of results so it may be something wrong with my input data in iDEA