Can DNA Methylation data of different samples from Human Methylation 27 and 450 platforms be combined using common set of 25,978 CpG sites?

Question

[khanaparajita]

I am studying the Kidney Renal Clear Cell Carcinoma (TCGA-KIRC) data set of TCGA. The DNA methylation datatype has 200 samples from Human Methylation 27 platform and 300 samples from Human Methylation 450 platform. These two platforms have methylation beta values for a common set of 25,978 CpG sites. Can I simply combine the samples from the two platforms using only the common set of 25,978 features?

Answer 1

[noorpratap.singh]

Yes, simple combining can be done. However I would suggest use existing implementations in packages such as minfi or MethylMix. In minfi there exists a function called 'combineArrays'.

[khanaparajita]

Should the data matrices corresponding to the different platforms be given as input to the combineArrays function?

[khanaparajita]

Can you suggest a reference where simple combination has been done?

[noorpratap.singh]

I apparently don't have a reference but coming to the first question the data matrices should be converted into an RGChannelSet, a MethylSet, a RatioSet, a GemomicMethylSet or a GenomicRatioSet. These terms are clearly explained here https://www.bioconductor.org/help/course-materials/2015/BioC2015/methylation450k.html.

This reference might also help Tcga Illumina Methylation Combining 27K And 450K