Skip to main content
. 2016 Oct 6;6:34838. doi: 10.1038/srep34838

Table 3. Performance (%) comparison on the partial- and full-length testing dataset.

    Partial-length testing dataset
Full-length testing dataset
CPAT CNCI PLEK lncScore CPAT CNCI PLEK lncScore
Human Accuracy 84.03 80.51 63.14 89.12 94.41 92.20 90.61 95.21
Sensitivity 76.19 65.40 31.76 84.15 94.97 89.00 85.96 95.56
PPV 92.12 96.46 94.83 94.61 95.46 98.16 98.62 96.64
Specificity 92.75 97.33 98.07 94.67 92.75 97.33 98.07 94.67
NPV 77.78 71.65 56.36 84.29 92.00 84.64 81.31 92.99
MCC 69.41 65.36 39.07 78.85 87.59 84.55 81.96 89.93
Mouse Accuracy 79.04 76.47 50.07 89.92 94.65 92.83 83.67 96.46
Sensitivity 72.88 69.24 35.34 88.39 94.19 91.56 81.17 97.35
PPV 97.97 98.05 90.91 97.61 98.39 98.48 95.75 97.78
Specificity 95.88 96.23 90.35 94.08 95.88 96.23 90.35 94.08
NPV 56.40 63.37 33.82 74.78 86.05 80.98 64.18 92.99
MCC 61.15 58.02 25.21 77.27 84.21 83.52 65.47 91.10

The default cutoff of CPAT, PLEK, and lncScore is shown in Fig. 4, and the default cutoff of CNCI is 0.