Skip to main content
. 2020 Nov 5;18:3335–3343. doi: 10.1016/j.csbj.2020.10.022

Table 3.

Evaluating inductive performance with supervised GRN inferring methods on balanced datasets. Feature E is the explicit expression features and G is graph embedding.

Methods Features E. coli
S. cerevisiae
Note
Accuracy Precision Recall MCC Accuracy Precision Recall MCC
SVM E 0.621 ± 0.000 0.628 ± 0.000 0.594 ± 0.000 0.242 ± 0.000 0.505 ± 0.000 0.557 ± 0.000 0.056 ± 0.000 0.026 ± 0.000 Baseline
G + E 0.704 ± 0.027 0.761 ± 0.009 0.596 ± 0.092 0.420 ± 0.045 0.643 ± 0.000 0.941 ± 0.001 0.304 ± 0.000 0.387 ± 0.001 Enclosed Graph + SVM
RF E 0.568 ± 0.000 0.595 ± 0.000 0.423 ± 0.000 0.141 ± 0.000 0.507 ± 0.000 0.520 ± 0.000 0.186 ± 0.000 0.019 ± 0.000 Baseline
G + E 0.635 ± 0.031 0.807 ± 0.057 0.359 ± 0.070 0.326 ± 0.061 0.658 ± 0.004 0.848 ± 0.012 0.384 ± 0.007 0.377 ± 0.009 Enclosed Graph + RF
GRGNN_PC (hop0) E 0.653 ± 0.001 0.652 ± 0.001 0.726 ± 0.153 0.306 ± 0.001 0.537 ± 0.000 0.674 ± 0.001 0.145 ± 0.000 0.121 ± 0.001
G + E 0.670 ± 0.150 0.677 ± 0.160 0.776 ± 0.134 0.352 ± 0.286 0.630 ± 0.072 0.777 ± 0.155 0.492 ± 0.290 0.306 ± 0.171
GRGNN_PC (hop1) E 0.586 ± 0.007 0.580 ± 0.007 0.625 ± 0.009 0.173 ± 0.014 0.566 ± 0.000 0.662 ± 0.002 0.395 ± 0.280 0.164 ± 0.000
G + E 0.696 ± 0.078 0.677 ± 0.062 0.773 ± 0.100 0.395 ± 0.160 0.655 ± 0.059 0.746 ± 0.121 0.518 ± 0.078 0.343 ± 0.115
GRGNN_MI (hop0) E 0.614 ± 0.002 0.581 ± 0.001 0.810 ± 0.025 0.251 ± 0.003 0.536 ± 0.000 0.678 ± 0.000 0.136 ± 0.001 0.119 ± 0.000
G + E 0.820 ± 0.008 0.874 ± 0.015 0.741 ± 0.034 0.647 ± 0.011 0.632 ± 0.175 0.866 ± 0.269 0.396 ± 0.070 0.321 ± 0.424
GRGNN_MI (hop1) E 0.652 ± 0.003 0.635 ± 0.002 0.718 ± 0.019 0.306 ± 0.006 0.534 ± 0.001 0.571 ± 0.001 0.326 ± 0.117 0.079 ± 0.002
G + E 0.767 ± 0.068 0.744 ± 0.077 0.847 ± 0.025 0.540 ± 0.134 0.566 ± 0.202 0.695 ± 0.283 0.579 ± 0.149 0.150 ± 0.453
GRGNN_EN (hop0) E 0.643 ± 0.000 0.619 ± 0.001 0.743 ± 0.002 0.293 ± 0.002 0.537 ± 0.000 0.676 ± 0.001 0.141 ± 0.000 0.120 ± 0.000
G + E 0.771 ± 0.100 0.766 ± 0.141 0.862 ± 0.076 0.568 ± 0.187 0.662 ± 0.090 0.818 ± 0.221 0.568 ± 0.229 0.388 ± 0.195 Baseline Compared
GRGNN_EN (hop1) E 0.656 ± 0.000 0.637 ± 0.000 0.730 ± 0.000 0.318 ± 0.003 0.570 ± 0.000 0.630 ± 0.002 0.340 ± 0.002 0.158 ± 0.001
G + E 0.809 ± 0.033 0.743 ± 0.069 0.853 ± 0.112 0.564 ± 0.153 0.684 ± 0.056 0.770 ± 0.147 0.574 ± 0.083 0.393 ± 0.135 Proposed Method