Skip to main content

Advertisement

Table 2 Comparison with state-of-the-art approaches in terms of AUPR

From: A unified solution for different scenarios of predicting drug-target interactions via triple matrix factorization

  NetLapRLS WNN-GIP RLScore KBMF2K CMF NRLMF TMF
S1-CV
 EN 0.789 ± 0.005 0.706 ± 0.017 0.828 ± 0.011 0.654 ± 0.008 0.877 ± 0.005 0.892 ± 0.006 0.952 ± 0.002
 IC 0.837 ± 0.009 0.717 ± 0.020 0.769 ± 0.015 0.771 ± 0.009 0.923 ± 0.006 0.906 ± 0.008 0.952 ± 0.002
 GPCR 0.616 ± 0.015 0.520 ± 0.021 0.625 ± 0.012 0.578 ± 0.018 0.745 ± 0.013 0.749 ± 0.015 0.844 ± 0.006
 NR 0.465 ± 0.044 0.589 ± 0.034 0.526 ± 0.045 0.534 ± 0.050 0.584 ± 0.042 0.728 ± 0.041 0.811 ± 0.035
S2-CV
 EN 0.123 ± 0.009 0.278 ± 0.037 0.313 ± 0.031 0.263 ± 0.033 0.229 ± 0.020 0.358 ± 0.040 0.438 ± 0.016
 IC 0.200 ± 0.026 0.258 ± 0.032 0.300 ± 0.020 0.308 ± 0.038 0.286 ± 0.030 0.344 ± 0.033 0.376 ± 0.017
 GPCR 0.229 ± 0.017 0.295 ± 0.025 0.368 ± 0.025 0.366 ± 0.024 0.365 ± 0.022 0.364 ± 0.023 0.428 ± 0.011
 NR 0.417 ± 0.048 0.504 ± 0.056 0.500 ± 0.058 0.477 ± 0.049 0.488 ± 0.050 0.545 ± 0.054 0.541 ± 0.033
S3-CV
 EN 0.669 ± 0.021 0.566 ± 0.038 0.794 ± 0.021 0.565 ± 0.023 0.698 ± 0.021 0.812 ± 0.018 0.866 ± 0.007
 IC 0.737 ± 0.020 0.696 ± 0.035 0.781 ± 0.026 0.677 ± 0.021 0.620 ± 0.027 0.785 ± 0.028 0.853 ± 0.008
 GPCR 0.334 ± 0.025 0.550 ± 0.047 0.533 ± 0.051 0.516 ± 0.045 0.433 ± 0.028 0.556 ± 0.038 0.677 ± 0.028
 NR 0.449 ± 0.074 0.531 ± 0.073 0.433 ± 0.079 0.324 ± 0.071 0.400 ± 0.077 0.449 ± 0.079 0.675 ± 0.062
S4-CV
 EN 0.238 ± 0.018 0.211 ± 0.020 0.265 ± 0.023
 IC 0.187 ± 0.020 0.232 ± 0.011 0.251 ± 0.014
 GPCR 0.208 ± 0.017 0.111 ± 0.034 0.231 ± 0.021
 NR 0.191 ± 0.051 0.231 ± 0.040 0.239 ± 0.037
  1. In S1, S2, S3, the results generated by former approaches were reported by [16]. The best results in each benchmark dataset under four kinds of CVs are highlighted in bold face and the second-best results are underlined