Table 2

AUCs and t-values obtained in the cross-validation experiment.

Model

The ratio of training to test data

3:1

1:1

1:3


3MAM(CG+CC+GG)

75.1

74.0

72.9

2MAM(CG+CC)

75.0(0.09)

73.7(2.36)

71.4(15.4)

2MAM(CG+GG)

72.4(26.2)

70.8(23.7)

68.4(47.0)

AM(CG)

73.3(15.1)

70.0(23.5)

64.7(65.7)


After training 3MAM with all three types of co-occurrence data, we computed the likelihood of all other cancer-gene pairs that are unknown in the OMIM. For each type of cancer, we present the top specific implicit gene in the Table 3. One interesting result is the top implicit associated gene specific to the prostate cancer, KLK10, which was already verified by Bharaj et al [3].

Zhu et al. BMC Systems Biology 2007 1(Suppl 1):P16   doi:10.1186/1752-0509-1-S1-P16

Open Data