I have a differntially expressed geneset. Some genes have an specific element in their promoters. I want to perform a hypergeometric test to see whether genes that have this element of their promoters are more likely to be differentially expressed than genes that do not have this element. I found that below one should be the relevent R function.
phyper(success-in-sample-1, success-in-bkgd, failure-in-bkgd, sample-size, lower.tail= FALSE);
What are the relevent parameters of my dataset. I can extract all the information from the dataset but dont understand what are they.
The hypergeometric distribution is a multivariate statistics, I think you don't need it if you have only 2 categories. in that case you can use the binomial distribution. Then you can see the whole thing in the following way. You have a set of black and white balls. Black are the balls with the specific motif in their promoter. You have a certain number of balls, sampled from the larger set, and they are both black and white (this is your set of differentially expressed genes). The question is: is the fraction of black balls in my set within or outside the range expected from the fraction of black balls in the genome? Or put it differently: what is the probability of getting as many or more black balls as there are in my set starting from the whole set of balls?
check wikipedia about binomial.
M
Depending on your skill level, a basic test could be Fisher's exact test or chi square test for independence. You can use binomial or regression based approaches if you are confident enough with those.
- Badges
- Science topic
- Similar topics
- Biological Science
- Genetics
- Gene Expression