Merging BS-Seq data from replicates

Question

[vipin.singh.cu]

Hi,

I have BS Seq data from three replicates and I want to merge them into one for further comparisons - what is the statistically most sound way of merging them ?

[Devon Ryan]

The most statistically sound way is to not merge them at all. What sort of comparisons do you intend to do after merging them?

[vipin.singh.cu]

Thanks …

So we have three data sets from the WT and mutant parents and we also have data from 121 recombinant inbred lines arising from the cross between the WT and mutant parent. We wish to compare the methylation score (ratio of methylated Cs /Total Cs) at each position in the parents and the inbred line offsprings And so I wish to have a single reference value for the WT and mutant Parents.

[vipin.singh.cu]

Thanks … So we have three data sets from the WT and mutant parents and we also have data from 121 recombinant inbred lines arising from the cross between the WT and mutant parent. We wish to compare the methylation score (ratio of methylated Cs /Total Cs) at each position in the parents and the inbred line offsprings And so I wish to have a single reference value for the WT and mutant Parents

[Devon Ryan]

I see no reason that you would want a single value for the parents in that case, using all of the samples would allow determining variance and using an actual test (e.g., with a beta binomial model).