jasminezhoulab / cfmethyl-seq Goto Github PK

Hi there,

Thanks for this inspirational idea and model to handle methylation data.

We have a couple of questions regarding data preprocessing. You have demo data for the model, but we don't find any for preprocessing. Can we have some here?

In our case, we have two groups, cancer and background. So I think samplesetfreq method should be used, correct? But when should we use "alpha.samplesetfreq.thresholds" and when should use "alpha.freq.thresholds"?

For the choice of hyper and hypo, we are guessing that it's case-dependent, based on our knowledge of the role of methylation in the disease/cancer, right?

The last question is about the thresholds, for example, in hyper.alpha.samplesetfreq.thresholds.n2.p10. How were the thresholds determined? Is there any reason that 2 is chosen for the negative class and 10 for the positive?

Thanks,
Panwen