Analysis Of Dna Methylation Associations Using 450K Data: Best Way To Control For Multiple Testing?

Question

[Luke]

Dear all. I am using Illumina Infinium 450k Methylation data in a number of human samples in an epidemiological analysis.

I would like to assess each site for its association with various outcomes, but am unsure as to the best method for controlling for multiple-testing in this context.

Methylation can be thought of as similar to SNP data, in that there is proximal correlation between CpG sites (akin to LD). The methods I've found for controlling for false-positives/negatives in SNP data seem to specific however, and I couldn't see an easy way to convert the method to continuous markers (as in methylation data).

Suggestions would be much appreciated - preferably with published methods.