Table 2.

Performance of the proposed module triad method and other network-based target prioritization methods on recovery of known approved targets in psoriasis, Parkinson’s disease and Alzheimer’s disease.

	Module triad prioritization			Reference prioritization methods
Performance metric	Combined (sel. & prox.)	Selectivity to TM	Proximity to GM	Local radiality	Random walk	node2vec
Psoriasis
Citeline, $\mathbf{31}$ approved targets
AUC	$\mathbf{0.73}$	0.71	0.65	0.70	0.72	0.48
AUPR	0.0039	0.0038	0.0032	0.0049	$\mathbf{0.0166}$	0.0068
Precision@500	0.002	0.006	0.004	0.004	$\mathbf{0.010}$	0.006
Recall@500	0.0322	0.0968	0.0645	0.0645	$\mathbf{0.1613}$	0.0968
Open access databases (union), $\mathbf{99}$ approved targets
AUC	0.72	$\mathbf{0.73}$	0.63	0.63	0.62	0.48
AUPR	$\mathbf{0.0143}$	0.0139	0.0090	0.0114	0.0132	0.0078
Precision@500	$\mathbf{0.024}$	0.020	0.014	0.012	0.022	0.014
Recall@500	$\mathbf{0.1212}$	0.1010	0.0707	0.0606	0.1111	0.0707
Open access databases (at least two), $\mathbf{25}$ approved targets
AUC	0.77	0.75	0.69	$\mathbf{0.79}$	0.73	0.40
AUPR	0.0041	0.0036	0.0031	$\mathbf{0.0134}$	0.0080	0.0016
Precision@500	0.004	0.004	0.004	0.008	$\mathbf{0.012}$	0.002
Recall@500	0.08	0.08	0.08	0.16	$\mathbf{0.24}$	0.04
$\underset{\_}{\mathit{P}\mathit{a}\mathit{r}\mathit{k}\mathit{i}\mathit{n}\mathit{s}\mathit{o}{\mathit{n}}^{\mathbf{\prime }}\mathit{s}\mathit{d}\mathit{i}\mathit{s}\mathit{e}\mathit{a}\mathit{s}\mathit{e}}$
Citeline, $\mathbf{13}$ approved targets
AUC	0.79	$\mathbf{0.82}$	0.65	0.55	0.59	0.51
AUPR	$\mathbf{0.0047}$	0.0046	0.0023	0.0009	0.0009	0.0007
Precision@500	$\mathbf{0.008}$	0.006	0.004	0.002	0.0	0.0
Recall@500	$\mathbf{0.3077}$	0.2308	0.1538	0.0769	0.0	0.0
Open access databases (union), $\mathbf{65}$ approved targets
AUC	$\mathbf{0.79}$	0.78	0.68	0.46	0.50	0.51
AUPR	$\mathbf{0.0185}$	0.0167	0.0115	0.0031	0.0035	0.0033
Precision@500	$\mathbf{0.030}$	$\mathbf{0.030}$	0.018	0.006	0.002	0.0
Recall@500	$\mathbf{0.2308}$	$\mathbf{0.2308}$	0.1385	0.0462	0.0154	0.0
Open access databases (at least two), $\mathbf{34}$ approved targets
AUC	0.85	$\mathbf{0.86}$	0.69	0.46	0.49	0.49
AUPR	$\mathbf{0.0138}$	0.0128	0.0074	0.0020	0.0020	0.0017
Precision@500	$\mathbf{0.020}$	$\mathbf{0.020}$	0.010	0.006	0.002	0.0
Recall@500	$\mathbf{0.2941}$	$\mathbf{0.2941}$	0.1471	0.0882	0.0294	0.0
$\underset{\_}{\mathit{A}\mathit{l}\mathit{z}\mathit{h}\mathit{e}\mathit{i}\mathit{m}\mathit{e}{\mathit{r}}^{\mathbf{\prime }}\mathit{s}\mathit{d}\mathit{i}\mathit{s}\mathit{e}\mathit{a}\mathit{s}\mathit{e}}$
Citeline, $\mathbf{9}$ approved targets
AUC	0.84	$\mathbf{0.85}$	0.69	0.61	0.59	0.36
AUPR	0.0048	0.0081	0.0015	0.0068	$\mathbf{0.0376}$	0.0031
Precision@500	$\mathbf{0.006}$	0.004	0.0	0.002	0.002	0.002
Recall@500	$\mathbf{0.3333}$	0.2222	0.0	0.1111	0.1111	0.1111
Open access databases (union), $\mathbf{78}$ approved targets
AUC	$\mathbf{0.72}$	0.65	0.69	0.53	0.50	0.46
AUPR	0.0125	0.0081	$\mathbf{0.0150}$	0.0059	0.0104	0.0042
Precision@500	$\mathbf{0.020}$	0.008	0.018	0.008	0.010	0.004
Recall@500	$\mathbf{0.1282}$	0.0513	0.1154	0.0513	0.0641	0.0256
Open access databases (at least two), $\mathbf{36}$ approved targets
AUC	$\mathbf{0.67}$	0.62	0.63	0.49	0.48	0.47
AUPR	$\mathbf{0.0049}$	0.0044	0.0037	0.0020	0.0018	0.0018
Precision@500	$\mathbf{0.008}$	0.006	0.004	0.002	0.002	0.0
Recall@500	$\mathbf{0.1111}$	0.0833	0.0556	0.0278	0.0278	0.0

Known approved targets are derived from Citeline database, as well as the three open access databases. Best performing methods are highlighted in bold.