Hello! I need some help to analyse methylation 450k data. I have 16 families where in each family children, parent and grandparents have been analysed for 450k array. All the children have had a type of disease, but parents and grandparents are unaffected.
I want to dissect the diseases associated methylation changes, but the problem is methylation levels are also confounded by age. Here is the a 3D PCA plot.
The black ones are children with disease, red ones are parents and green ones are grandparents. Disease and generation are the first two principle components. Is there any way to adjust the age ? The problem is although Children are affected they are also younger, so in my case disease state and age are interacting...
I have added age as a continuous variable to adjust in the Combat function in SVA package, when I adjust batch effect together. The PCA plot is after age adjustment, but they are still perfectly labelled by generation, so Combat didn't seem to work.
Instead of jumping into ComBat, have you tried adding the Age term in your model when doing differential methylation? Does that improve your number of differentially methylated probes?