Hello all,

I'm having some trouble removing batch effects from my data.

I have a matrix of methylation beta values where the rows are probes and the columns are samples/patients. I additionally have a table indicating what platform was used for each sample/patient (27K or 450K Methylation Arrays).

I performed a PCA on the logit transformed beta values and it showed clear clustering based on platform.

I would like to adjust/remove the batch effect from my matrix of methylation beta values, but I have run intro trouble with all the packages I tried (sva, ChAMP, the ComBat R script).

My ultimate goal is to subtype/cluster the samples, so I don't have a variable of interest. All the patients belong to the same phenotype.

I would greatly appreciate any insight on how to remove the batch effect from the methylation platform.

Thanks!