Hi, I am trying to analyze data from Illumina Expression BeadChips.

So, imagine we have separate matrices for different sets of patients (in GEO (https://www.ncbi.nlm.nih.gov/geo), like GSE167093 case), that most likely were made on separate chips (because of great number of samples). I want to get just one big table with all the patients, but I tend to expect serious batch effects while just merging these tables together. Is there a way to overcome this issue (cross-beads normalization, something like this) or it is not a good idea and these matrices should be processed separately?