Methylation Data Interpretation and Dychotomization

Question

[luisa]

Hi! I am working with methylation data from the TCGA Database. I chose the file with the mean signal values among each gene... The data presented is beta-values, which range from 0 to 1, where 0 means the gene is not methylated and 1 that the gene is totally methylated, however there are values in between and I am trying to dycotomize my data into methylated/non-methylated... What is the best way to set a threshold to define my two groups?

Thanks in advance!

Answer 1

[suleymanvural]

Not a simple task. See the post here A: Interpreting Fractional Methylation Data

[Kevin Blighe]

Just to add: I recently set Beta=0.6 as a 'hard' threshold and our work has just been accepted for publication in OncoTarget.

There is no real answer, though. It's like fold change differences in expression studies.

[luisa]

I actually think that is the easiest way... I have also seen other papers that take that approach. However, as you say I have already come across various methods to do this... I think I will look at my data and set an empirical threshold... Thank you both!