I'm looking to carry out DEXseq (for differential exon usage) and Slueth (differential gene expression) analysis on some TCGA BAM files. As the BAM files can be from different centers and created in different sequencing batches I was looking to control for hidden batch effects using PEER (run on normalized expression obtained from Kallisto), and supplying the hidden factors identified.
Would this approach be justified? Also I would be grateful for a reference to how this can be done in the case of Slueth.
Thanking You, Vakul