A novel analysis strategy for integrating methylation and expression data reveals core pathways for thyroid cancer aetiology

Table 9 Validating the proposed framework with a new dataset.

	Training Dataset (>15% Methylation Change, Significances Combined)	Test Dataset (>15% Methylation Change, Significances Combined)
KEGG TERMs	Pooled	Pooled
MAPK Signalling	6	11
ECM Receptor	3	1
ErbB Signalling	10	16
NF-KB Signalling	22	25
Wnt-β-Catenin Signalling	72	80
VEGF Signalling	30	69
Thyroid Cancer	45	68
Adherens Junction	14	19
p53 Signalling	54	17
TGF-beta Signalling	13	7
Notch Signalling	12	83
GnRH Signalling	56	15
Neurotrophin Signalling	17	12
Focal Adhesion	2	2
Transcr. Misregulation	51	-
Apoptosis	4	20
Pathways in Cancer	1	4
Toll-like receptor signalling pathway	8	-
Pentose-phosphate pathway	18	75

Comparison between the training and test dataset. For the training dataset, optimal results were obtained using Model 7. When the same analysis procedure of training dataset is applied to the test dataset (30 samples), similar results were obtained. There were 12 KEGG functional annotation terms for the training dataset whereas this number was 11 for the test dataset, which shows that our proposed framework is also applicable to independent datasets.

ISSN: 1471-2164