Dear edgeR support team,

In the tutorial its mentioned that Batch effects can be removed or corrected by using the additive model formula. How can I visualise the removal of batch effect on the clustering of the samples using edgeR.

Thanks and Regards,

Candida

I recall one thread in particular about removing batch effects from count data that might be of interest, which is RNA-Seq, generate batch-free count matrix 

A: RNA-Seq, generate batch-free count matrix is what you're really interested in. You won't get count data back as it must first be log2 transformed (with some smoothing via the prior.count parameter), but perhaps you can do the log2 transform before removing batch effects and after to see what changes via MDS plots, heatmaps, etc.

When doing your differential expression analysis, though, you'll just want to work with the original counts and add the batch effects in your design matrix so that they are properly controlled for.

To add to Steve's reply, here is another previous thread with the same advice: Remove batch effect in small RNASeq study (SVA or others?)

If you are trying to visualize the effects of a batch effect in a linear model, then you should specify the design matrix (minus the batch factor) as well as the batch factor when calling removeBatchEffect().