I am attempting to use edgeR see if the relative abundance of a pair of genomic features changes between different tissue samples. I have samples from 46 different tissues and each sample has 3 bioReps. For now I am using the exactTest method on a DGEList with 6 columns and 46 rows. Each column is a bioRep from one of the two features, and each row is a tissue. I've made up an example below.
I am hoping to be able to see which tissues have a ratio of feature 1 to feature 2 that deviate from the average ratio. So assuming that the average ratio between features 1 and 2 is 1:1 I would expect tissue 1 to show a bias towards feature 2 and tissue 3 to show a bias towards feature 1.
I would like to use edgeR to do this test because it allows for multiple bioReps, but I was wondering if this is a valid input for the exact test because the raw counts are all coming from different samples. For now I am not doing any kind of normalization, but I am not sure if I should. Does anyone on here know if this is an appropriate use of the exactTest method in edgeR? Is there is another R package that is designed to do what I want?
I am hoping to be able to see which tissues have a ratio of feature 1 to feature 2 that deviate from the average ratio. So assuming that the average ratio between features 1 and 2 is 1:1 I would expect tissue 1 to show a bias towards feature 2 and tissue 3 to show a bias towards feature 1.
I would like to use edgeR to do this test because it allows for multiple bioReps, but I was wondering if this is a valid input for the exact test because the raw counts are all coming from different samples. For now I am not doing any kind of normalization, but I am not sure if I should. Does anyone on here know if this is an appropriate use of the exactTest method in edgeR? Is there is another R package that is designed to do what I want?