Need help in DNA methylation data processing

Question

[BioMed]

Dear all,

I'm new to DNA methylation data. I got into trouble because I'm not sure about any typical approaches (and corresponding software) in DNA methylation data processing. Here is what I have known:

• Get raw data.
• Perform data normalization.
• Conduct data alignment (BRAT-bw alignment tool).
• Define unmethylated probles and methylated probes criteria (<0.3 and >=0.3).
• Check for data quality control: Exclude data of the probes that has more than 10% or 20% missing data.
• Consider missing rate as well as its distribution (random or not). I'm not sure about this step.

Besides, please give me your opinions about the use of beta-value and M-value. which is better if I want to define the association between DNA methylation heterogeneity and the risk of a particular disease?

Another problem that is related to network/pathway analysis, could I treat the DNA methylation data similarly to gene expression data (using log(10) of P-value). Would KEGG annotation work well?

Any advice and references are appreciated. Thank you very much.

Answer 1

[igor]

It sounds like you have methylation array data, since you mention probes. I will assume you are using the common 450K arrays. Here is a good tutorial for analyzing 450K data: https://www.bioconductor.org/help/course-materials/2015/BioC2015/methylation450k.html