Table 2.

Common genes with AUC > 0.94: (90% specificity is achieved with more than 90% sensitivity) for D1, D2 and D3.

Datasets	Potential methylation biomarker genes
1. D1 and D2	ESR1, SLC6A15, HSU79303, SFRP2, MAL, C2orf32, ATP8B2, EOMES, FLI1, GAS7, LRRC4, FOXE1, ITGA4, BASP1, FBN2, COL4A1
2. D2 and D3	RLN3R1, RIC3
3. D1 and D3	ALX4 (4), DLX5 (6), MSX1 (6), ADHFE1 (6), CBLN4 (6), GCM2 (6), VGCNL1 (6), SLIT2 (4), WT1 (5,6), NPY (6), HCN1 (6,4), BNC1 (6), SLC18A3 (6), GHSR (6), SFRP1 (6), DEPDC2 (4), FLJ46831 (6), NTRK3 (4,6), SCARF2 (6,4), GPR75 (6), GPC6 (6), POU3F1 (4,6), KHDRBS2 (6), ZNF677 (6), VIPR2 (4,6), LCAT(6)
4. D1, D2 and D3	TRH (6), EYA4 (4), GRIN2A (4), MDFI (4), MSC (6), TFPI2 (4), RYR2 (6), PCDHGC4 (5,4), IRF4 (4), KCNQ5 (6), FLJ25477 (6).

Lists 3 and 4 should be noted, particularly, as the former gives the biomarker gene list overlap for the two larger datasets, while List 4 gives the overlap of these two with the smaller dataset D2. The corresponding gene cluster for D1 and D3 is given in brackets: (a single figure means that the gene clusters correspond).