Skip to main content
. 2014 Aug 22;30(17):i572–i578. doi: 10.1093/bioinformatics/btu466

Table 1.

Performance of the methods on different tumor types

Tumor Method Q2 PPV TPR NPV TNR C AUC
COAD CRank 0.92 0.97 0.86 0.87 0.97 0.84 0.94
CLow 0.72 0.78 0.66 0.72 0.78 0.47 0.79
CDiff 0.76 0.78 0.77 0.80 0.74 0.55 0.83
LUAD CRank 0.97 0.97 0.96 0.96 0.97 0.93 0.99
CLow 0.84 0.88 0.79 0.81 0.89 0.69 0.91
CDiff 0.89 0.89 0.90 0.90 0.88 0.79 0.96
PRAD CRank 0.91 0.92 0.91 0.91 0.92 0.83 0.97
CLow 0.70 0.66 0.75 0.72 0.74 0.43 0.77
CDiff 0.79 0.83 0.74 0.77 0.83 0.58 0.87

Note: Performance of ContrastRank, ContrastLow and ContastDiff (respectively, CRank, CLow and CDiff) calculated using an average number 239, 494 and 127 genes with score >3 respectively for COAD, LUAD and PRAD. Q2, overall accuracy; PPV and NPV, positive and negative predicted values; TPR and TNR, true positive and negative rates; MCC, Matthew’s correlation; AUC = area under the (ROC) curve.