make the ratio cut off to 0.75. That would give a more specific result, statistically, since epigenetic modification is very versatile. It may not always the same patter post treatment.

Hello Edoardo, you should initially filter-in all the predicted CpG islands. You can retrieve a CpG track from the genome browser (http://genome-euro.ucsc.edu/). By using this approach you can easily identify all the CpGs that are differentially methylated and then associate the corresponding gene, if present.

You can quantitate CpG ratio in 3 D with Cork and Lembark's W-curve: see Plos One HIV-1 Genomic Comparisons with the W-curve, 2010 , Feb.