Table 2 Performance comparisons of the different structural descriptor-based methods and DPDR-CPI using 638 endpoints of ICD-9 disease indications on the independent validation data.

From: DPDR-CPI, a server that predicts Drug Positioning and Drug Repositioning via Chemical-Protein Interactome

Metric	Method	Accuracy	Precision	Sensitivity	Specificity	AUROC	AUPR
global	LR-ECFP6	0.801	0.040	0.355	0.811	0.651	0.040
	LR-E-State	0.792	0.033	0.305	0.802	0.602	0.030
	LR-FCFP6	0.904	0.050	0.193	0.919	0.670	0.051
	LR-FP4	0.870	0.045	0.247	0.883	0.633	0.038
	LR-KR	0.897	0.045	0.190	0.912	0.660	0.048
	LR-MACCS	0.881	0.047	0.235	0.896	0.649	0.039
	LR-PubChem	0.866	0.054	0.318	0.878	0.704	0.050
	DPDR-CPI	0.964	0.192	0.203	0.981	0.764	0.118
drug-centric (628)	LR-ECFP6	0.906 ± 0.140	0.316 ± 0.303	0.490 ± 0.267	0.916 ± 0.144	0.783 ± 0.154	0.135 ± 0.161
	LR-E-State	0.888 ± 0.153	0.235 ± 0.245	0.465 ± 0.273	0.898 ± 0.158	0.744 ± 0.150	0.085 ± 0.087
	LR-FCFP6	0.902 ± 0.147	0.298 ± 0.292	0.497 ± 0.267	0.911 ± 0.150	0.781 ± 0.153	0.132 ± 0.156
	LR-FP4	0.897 ± 0.145	0.258 ± 0.261	0.472 ± 0.266	0.907 ± 0.149	0.761 ± 0.146	0.103 ± 0.116
	LR-KR	0.903 ± 0.139	0.289 ± 0.290	0.498 ± 0.265	0.912 ± 0.142	0.778 ± 0.152	0.129 ± 0.156
	LR-MACCS	0.895 ± 0.148	0.271 ± 0.276	0.478 ± 0.269	0.905 ± 0.152	0.762 ± 0.152	0.108 ± 0.124
	LR-PubChem	0.897 ± 0.156	0.293 ± 0.294	0.486 ± 0.265	0.907 ± 0.160	0.766 ± 0.160	0.125 ± 0.146
	DPDR-CPI	0.893 ± 0.150	0.273 ± 0.282	0.511 ± 0.271	0.902 ± 0.154	0.775 ± 0.156	0.128 ± 0.163
disease-centric (638)	LR-ECFP6	0.668 ± 0.268	0.061 ± 0.075	0.559 ± 0.305	0.671 ± 0.278	0.563 ± 0.138	0.032 ± 0.036
	LR-E-State	0.596 ± 0.323	0.059 ± 0.097	0.593 ± 0.338	0.596 ± 0.336	0.504 ± 0.150	0.026 ± 0.026
	LR-FCFP6	0.711 ± 0.242	0.077 ± 0.105	0.529 ± 0.293	0.716 ± 0.250	0.583 ± 0.136	0.036 ± 0.044
	LR-FP4	0.629 ± 0.307	0.064 ± 0.093	0.575 ± 0.326	0.632 ± 0.318	0.524 ± 0.148	0.030 ± 0.035
	LR-KR	0.690 ± 0.259	0.109 ± 0.212	0.553 ± 0.308	0.695 ± 0.268	0.571 ± 0.133	0.035 ± 0.048
	LR-MACCS	0.659 ± 0.286	0.061 ± 0.083	0.549 ± 0.310	0.663 ± 0.296	0.536 ± 0.145	0.030 ± 0.034
	LR-PubChem	0.746 ± 0.231	0.075 ± 0.090	0.523 ± 0.283	0.752 ± 0.239	0.609 ± 0.144	0.039 ± 0.046
	DPDR-CPI	0.888 ± 0.173	0.258 ± 0.261	0.388 ± 0.234	0.899 ± 0.179	0.682 ± 0.148	0.088 ± 0.091

Three types of metrics, including global, drug-centric and disease-centric metrics, were used.

Back to article page

Table 2 Performance comparisons of the different structural descriptor-based methods and DPDR-CPI using 638 endpoints of ICD-9 disease indications on the independent validation data.

Search

Quick links