I am strugelling with TCGA to get an accurate clustering. I have tried to include many covariates and batch factors to my model using voom and removeBatchEffect(). The problem is my PCA is not satisfactory for me as there is always some points that cluster aways from similar condition and different conditions that overlaps.
Most of the studies I have read so far have not reported the clustering results. They said they accounted for the batch.
So, what is your experiences? Does any one succesfully removed the batch effect and clustered the data?