Table 4.
Summary of performance of QSAR models for predicting the oligomeric state of FPs (95% homologous sequence reduction) using the J48 algorithm
Descriptors | Training set | Tenfold CV set | External set | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Ac (%) | Sn (%) | Sp (%) | MCC | Ac (%) | Sn (%) | Sp (%) | MCC | Ac (%) | Sn (%) | Sp (%) | MCC | |
AAC/DPC/TPC | 97.54 ± 1.19 | 99.25 ± 0.90 | 94.85 ± 2.50 | 0.95 ± 0.03 | 72.13 ± 4.18 | 79.83 ± 3.66 | 61.03 ± 5.34 | 0.42 ± 0.09 | 72.89 ± 7.08 | 79.85 ± 6.92 | 64.16 ± 11.20 | 0.43 ± 0.15 |
AC | 98.35 ± 0.87 | 99.31 ± 0.92 | 96.81 ± 1.95 | 0.97 ± 0.02 | 70.71 ± 4.45 | 77.73 ± 3.63 | 59.80 ± 6.10 | 0.38 ± 0.09 | 70.30 ± 8.55 | 77.40 ± 7.91 | 60.99 ± 13.19 | 0.38 ± 0.18 |
CTD | 97.97 ± 1.06 | 98.33 ± 1.40 | 97.50 ± 1.95 | 0.96 ± 0.02 | 69.40 ± 4.95 | 75.24 ± 4.39 | 60.62 ± 6.33 | 0.39 ± 0.10 | 70.18 ± 7.79 | 75.54 ± 7.39 | 63.17 ± 12.39 | 0.38 ± 0.17 |
Ctriad | 96.62 ± 1.33 | 98.07 ± 1.52 | 94.35 ± 2.89 | 0.93 ± 0.03 | 68.64 ± 5.99 | 76.28 ± 4.49 | 57.20 ± 8.12 | 0.34 ± 0.12 | 71.26 ± 8.36 | 78.04 ± 7.20 | 62.51 ± 12.24 | 0.40 ± 0.17 |
QSO | 98.10 ± 1.08 | 98.55 ± 1.25 | 97.42 ± 2.33 | 0.96 ± 0.02 | 68.98 ± 4.21 | 76.15 ± 3.45 | 57.59 ± 5.63 | 0.34 ± 0.09 | 69.93 ± 6.90 | 77.19 ± 5.75 | 60.30 ± 11.15 | 0.37 ± 0.14 |
PseAAC | 98.24 ± 0.92 | 98.38 ± 1.30 | 98.07 ± 1.71 | 0.96 ± 0.02 | 69.39 ± 4.97 | 76.34 ± 3.98 | 58.20 ± 6.67 | 0.35 ± 0.10 | 69.67 ± 8.03 | 76.92 ± 7.01 | 59.53 ± 10.36 | 0.36 ± 0.17 |