Hi,

I have been working on an RNA-Seq dataset on which I applied RUVSeq (ruvr method) at the gene-count level to account for some unexpected variance before running DESeq2. I am also working on using DEXSeq to look at differential exon usage. Optimally, I would be able to also apply RUVSeq to the binned exon counts, so that both gene and exon level counts would have been processed similarly. However, I'm not sure if it's appropriate to do this at the exon level- any ideas?

Both of these packages (RUVSeq and of course DESeq2) are BioConductor packages.

Thanks!

Hi Adam,

sorry for the late response. Provided that you can identify negative control exons, I don't see any problem with applying RUVSeq to exon data. You should be able to pass the factors of unwanted variation to DEXSeq in the design matrix, in a similar way as you would do with DESeq.

We had some good results by just assuming that all the exon bins of the negative control genes were not DE in this paper: 

http://www.sciencedirect.com/science/article/pii/S1074742716301186

I hope this helps.

Davide