The wig file you loaded is their final track deposited in GEO, which was generated with a 20-bp sliding window across the genome, for purpose of consistency and easy comparison. So each bar would cover a range of sequence in multiples of 20 bps, which may contain one or more CpG sites of interest. You could kindly contact the publisher for additions about the dataset if they were not found in GEO. Hope this helps, thanks!