Entering edit mode
8.5 years ago
CandiceChuDVM
★
2.5k
Hi all,
As a follow-up of this post, I am tempted to provide a list of genes (with no numeric values) to the http://www.pantherdb.org/ for statistical overrepresentation test.
- Any downside of taking this route?
- As far as I know, I can provide a list of genes with numeric values for statistical enrichment test. Is it better than overrepresentation test? If so, what kind of numerical values should I provide from DESeq2 output?
- In terms of providing "better" results (not just the convenience of using R), should I use other R packages for gene ontology analysis? For example, GOseq?
Thanks!
Were the gene lists split into up-/down- regulated for the submission?
That's the other question that I have been thinking. The related post is available here: "Separate enrichment analysis of pathways for up- and downregulated genes"
I use goseq for results from RNA-seq studies. It's similar as enrichment analysis, but it takes a length bias into account.
You don't need any values, only a list of significant genes (and a list of all genes in your experiment).
Hi b.nota,
By "a list of all genes in your experiment", do you mean "reference genome of the species used in the experiment"? I always use reference genome as my background without thinking. Maybe I should use "a list of all genes in your experiment" instead.
Hi, sorry for the late reply...
Yes, all genes from reference can be used. But alternatively you can also use all genes that are left after filtering in e.g., edgeR. Usually you filter out genes first, with no or low number of reads before differential analysis.