We have an RRBS dataset with 4 treatment groups, n=6 samples each group. I'd like to be able to do an ANOVA-style first stage analysis to look for evidence that treatment effects exist in the dataset as a whole, before moving forward into pairwise post-hoc analyses. We've been using Methylkit (beta binomial distribution) for pairwise analyses, but unless I'm mistaken it doesn't support multi-group comparisons. Is there another package that does? Or is this something we would have to develop on our own (as a GLM in R for example)? Advice appreciated!
-david
-david