This is a test version of Biostars. For the public version, visit https://www.biostars.org.
how to detect non CpG methylation probes ?

Hi, everyone :)

I want to ask methods about post title.

I think we can detect DMP or DMR using R packages minfi in infinium450k.

But, I couldn't get programs for detection of non CpG methylation probes using minfi.

Does anyone know about non CpG methylation probes detection ?

Thanks in advance.

dnamethylation methylation infinium450k microarray minfi r
• 2,092 views • link
written by takoyaki takoyaki

1 answer

The 450k array does measure non-CpG probes (around 3000 of them). These are identified by starting with "ch." in their probe IDs. You can check them in the annotation like this:

library(IlluminaHumanMethylation450kanno.ilmn12.hg19)
ann450k <- getAnnotation(IlluminaHumanMethylation450kanno.ilmn12.hg19)
ann450k.f <- ann450k[grepl("ch",ann450k$Name),]
head(ann450k.f)

AFAIK, these probes are not filtered out by minfi, they are treated equally as the CpG probes and retained after preprocessing, so you could filter your data to get them and use them to perform subsequent analyses.

Beware though, that many studies have stated that a large proportion of these non-CpG probes may be cross-reactive (example1, example2).

• 0 views • link
written by Papyrus Papyrus

Thank you for your helpful comments and informative suggestion.
I've confrim that there are probes starting with "ch" !

However, I couldn't confirmed whether that probe show CpA or CpT and so on.
How do you distinguish that methylation type ?

• 0 views • link
written by takoyaki takoyaki

Well, the annotation package actually contains the sequence of the targeted locus by each probe in the $Forward_Sequence variable. The targeted locus is indicated by square brackets, like this example:

GATGACAAGCAGTTTATTTGCACTTTTGTTAGGGCTGCGGTGACTTACATGTCTGGCAAG[CA]CTTCTTTGCAAAATGTGACAACTGCTGCCATGAAAATGCACTGGTTTTAATGACAATAAC

So you could generate a variable like this (following the previous code):

ann450k.f$locus <- stringr::str_split_fixed(ann450k.f$Forward_Sequence, "\\[|\\]",3)[,2]
table(ann450k.f$locus)
CA   CT 
3081   10

It looks like almost all of the probes are CpA.

• 0 views • link
written by Papyrus Papyrus

I missed the Forward_Sequence column,,,

But this is surprising for me because almost all of the CpH is CpA !

Thank you for your help.

• 0 views • link
written by takoyaki takoyaki

Log in to answer this question.

More posts like this