search

EpigenomeClock / MouseEpigeneticClock

GNU General Public License v2.0
12 stars 7 forks source link

How to deal with missing CpGs in RRBS and interprete the predicted DNAm age results #14

Open liuyang2006 opened 1 year ago

liuyang2006 commented 1 year ago

Dear author,

I am deeply intrigued by your epigenetic clock program. However, while running your program, I noticed some missing CpGs in my RRBS data. Could you advise if it's feasible to use your clock for predicting my data even with this issue?

After I proceeded with the results, having removed the missing CpGs, how should I interpret them? Are the results represented in weeks, months, or another time unit?

I appreciate your guidance on this matter.

Best, Yang

Bonder-MJ commented 1 year ago

Hi Yang,

Thanks for your interest. You can use several approaches when not all sites are covered. You can try to use the imputation approach also on this github page: https://github.com/EpigenomeClock/MouseEpigeneticClock/blob/master/toRun_Imputation.R

Alternatively you can use our new approach:https://github.com/EpigenomeClock/scEpiAge

The predictions that you get out ages in week.

Best, Marc

kwglam commented 5 months ago

Hi Marc,

I have tried running both toRun_Imputation.R and scEpiAge on methylation data from mouse liver samples. Both programs generated some very different values. For example, the predictedAges on 3 replicates using toRun_Imputation.R are 20.7, 24.9, and 20.9, whereas the predictedAges on the same samples using ScEpiAge (bulk) are 73, 47, and 119. Why is there such a big difference between programs? Are both results ages in week?

Also, I would like to confirm if the CpG sites that you used are 1-based CpG sites. Thanks a lot for your help and insights!!

Cheers, Gabriel