Prediction of CpG-island function: CpG clustering vs. sliding-window methods

BMC Genomics

Table 8 Number of unmethylated and differentially methylated CpG 'islets'.

Dataset	Methylation state*	Number of CpG islets	CpG 'islets' exclusively predicted by CpGcluster
HEP (12 tissues)**	Unmethylated	126	1
	Differentially methylated	26	8
Lister et al. 2009 (2 cell lines)***	Unmethylated	4,460	1,472
	Differentially methylated	373	295

*Unmethylated: average methylation ≤ 0.2; differentially methylated: average methylation <= 0.2 in at least one tissue & average methylation >= 0.8 in at least one other tissue.
**The methylation state of 246 CpG 'islets' from chromosomes 6, 20 and 22 was determined by using 3,168 individual CpG sites (HEP project). We only included CpGs which have been detected in at least 2 clones or in at least 6 different tissues.
***We used the sequence reads obtained by MethylC-Seq for two human cell lines [33], H1 human embryonic stem cells and IMR90 fetal lung fibroblasts, to get the average methylation level of single cytosines at both DNA strands for these two methylomes. All islands need more than 50% of its CpGs covered. Only cytosines covered by at least 10 reads were counted.

ISSN: 1471-2164