Dear community,

I'm fairly new to working with EWAS data and I currently have access to a dataset where the normalization was conducted using minfi::preprocessFunnorm with nPC = 30, which are principal components of the control probes (I know that this is much more than suggested). These steps were not performed by me.

However, in all later data analyses, I'm following a guideline from Lehne et al. (2015; CPACOR guidelines; https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4365767/), which state that the first 30 principal components of the control probes need to be set as covariates in the final EWAS model.

I was wondering if it really makes sense to control for the control probe PCs during functional normalization (see above) and to also (!) use these PCs as covariates in all later analyses?

Unfortunately, I cannot find anything in the rest of the internet, so all help would be greatly appreciated!

Thanks in advance and best wishes!

Martin