preprocessFunnorm vs preprocessNoob

[komalsrathi]

Hi,

I would like some clarity on two questions I have:

1. If I want to merge EPIC v1 and EPIC v2 arrays, does this methodology look ok

   • Read them separately, normalize them separately using preprocessFunnorm or preprocessNoob.
   • Format Probe IDs in EPIC v2 (they use Illumina ids by default)
   • Combine the resulting m-values, beta-values or cn-values from both arrays into single matrix by doing a full join.
   • This will of course result in NA values where probe values are not present. In such cases, do you have any recommendation for imputations?

2. Between the two normalizations: preprocessFunnorm and preprocessNoob, which one is recommended as I saw the preprocessFunnorm does Background and dye bias correction with noob as one of the first steps. In which cases would I prefer one over the other? The data we have consists of samples from various tumor types and normal tissues.

Thanks in advance,
Komal