Hi Kasper and Jim, I got DMRs on my local PC in reasonable execution time with this function call and basic output: dmrs.senzaXY <- bumphunter(myGenomicRatioSet.senzaXY, design=designMatrix, cutoff=0.99, cutoffQ=0.99, B=100, verbose=TRUE) [bumphunterEngine] Parallelizing using 2 workers/cores (backend: doParallelMC, version: 1.0.7). [bumphunterEngine] Computing coefficients. [bumphunterEngine] Performing 100 permutations. [bumphunterEngine] Computing marginal permutation p-values. [bumphunterEngine] cutoff: 0.99 [bumphunterEngine] Finding regions. [bumphunterEngine] Found 40696 bumps. [bumphunterEngine] Computing regions for each permutation. I don't know if 40696 is again a "crazy" number of putative bumps. The n samples of the current dataset is 10, probably this influences the results. I'm waiting for other samples of the next Illumina chip-array by my colleagues in such way to have a dataset with a doubled size, 20-24, and so execute the next bumphunter run. Thanks for the advice, Best Giovanni [[alternative HTML version deleted]]