differential methylation of illumina 450k data

Question

[Govardhan Anande]

Hi,

I am new to illumina 450k methylation arrays and I would like to study differential methylation between two samples. Any suggestion for tools/R packages for it. Output should have b-value as well.

Thanks in advance.

Answer 1

[Charles Warden]

1) COHCAP also directly uses beta values for differential methylation (and there is a table of beta values per island):

https://www.bioconductor.org/packages/release/bioc/html/COHCAP.html

It uses an annotation-based strategy, so you immediately get gene mappings but it doesn't identify boundaries of differentially methylated regions.

2) bumphunter (in minfi) is good option that that uses percent methylation to identify regions:

https://bioconductor.org/packages/release/bioc/html/minfi.html

It will provide boundaries for the differentially methylated regions, but it doesn't directly provide beta values in the result (although you can get beta values for each site using other minfi functions)

3) RnBeads also usually gives good results, but the p-value is based upon M-values. However, as long as you have access to delta-beta values, you can also compare delta-beta versus -log(p-value).

http://rnbeads.mpi-inf.mpg.de/