Hi
I just started a bioinformatics course, and a few days ago, I listened to the lecturer explain functional enrichment analysis, except it went through my head.
I was wondering if someone here could explain it nicely :-)
Chris
Perhaps you are referring to the situation where you do some type of analysis on high throughput data which results in a list of "interesting" genes. The typical scenario I'm thinking about is running some type of transcriptome analysis between two conditions of interest.
Let's further assume that this list of "interesting" (differentially expressed) genes has 100 members. Then what?
You want to ask if there is something "interesting" about this group of genes. I guess this is what "functional enrichment analysis" is referring to.
Do many of them seem to be involved in some type of pathway we know about? The Wnt signaling pathway, for instance? What if 10 of the genes in your list are in this pathway, is that more than you would expect by chance if you just fished out 100 genes at random?
To answer the question above, people typically do some gene ontology enrichment analysis.
You could also do something like GSEA, which doesn't require you to pick a cutoff that says this gene is interesting or not, but rather takes the entire list of genes in the experiment ranked by some metric (here, it would be the log-fold change between your two experimental conditions), then looks for enrichment of genes at the top of you list (or bottom) within different gene signatures that have been curated elsewhere.
If you google for "gene ontology analysis", or "gene set enrichment analysis" you'll get tons of information you can read through that will fill in the blanks of this answer, but this is the general idea.
Enirchment analysis is usually based on the hypergeometric distribution. If you wiki hypergeometric distribution the article explains it quite nicely but briefly, You have an urn filled with white balls and black balls.
The balls are all genes or other unit of analysis.
White balls are units that belong to the group you want to test for. Black balls are units that do not belong to that group.
Balls drawn are the number of significantly different units (says differentially expressed genes) White balls drawn are the number of units that belong to the group you are testing AND that were differentially expressed.
So you test what if the likely hood of drawing that many white balls from the urn in x number of draws when the total numbers of black and white balls are known. Basically it's a binomial distribution but WITHOUT replacement.
How you calculate that is up to you though, and is a whole other question! :)
Hope that helps!
http://stke.sciencemag.org/cgi/content/abstract/sigtrans;4/190/tr4
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
Hi Davy - I understand what gene expression is, but what does "differentially expressed" mean? Kind Regards Chris