I am using DEXSeq in a dataset with a couple of experimental conditions.

As I would do in DESeq I had planned to calculate normalization factors, dispersion and the GLM using the whole dataset and then calculate alternative splicing for selected contrasts only (e.g. condition2 versus condition1, condition4 versus condition3 ... ). This ought to be preferable compared to splitting up the dataset into pairwise subsets and calculating differences for those separately.

All the examples I've found so far are with two experimental conditions only.

How can you define contrasts in DEXSeq or would that be pointless in the first place for some reason?

Best,

Maik

Dear Maik,

If you want to compare only subsets of your experimental conditions, you need to subset your DEXSeqDataSet object such that it contains only the samples that you want to compare. For example, if you have condition1, condition2, condition3 and condition4, but you would like to compare condition1 vs condition2, you should leave only the samples with those conditions and rerun the DEXSeq pipeline on this subset.

Alejandro