Integration Strategy Is a Key Step in Network-Based Analysis and Dramatically Affects Network Topological Properties and Inferring Outcomes

Table 8

Performance of each network built by integration strategies for the discovery of drug targets based on topological properties.

Strategy	1N	Target	Node	Target ratio
Union	82	1984	14936	0.132833
Intersection	40	133	507	0.262327
2-Vote	79	1464	9548	0.153331
3-Vote	55	885	5558	0.15923
SVM	83	1974	14895	0.132528
Naive Bayes	83	1969	14840	0.132682
Random Tree	81	1941	14486	0.133991
J48	81	1945	14570	0.133493
Logistic Regression	85	1981	14890	0.133042
Random Forest	83	1972	14860	0.132705
Bayesian Networks	83	1973	14869	0.132692

Note: 1N indicates the number of targets included in the top 100 proteins. Target indicates the number of targets in the network. The target ratio reveals the percentage of targets in a network. The bold type indicates the maximum values in the 1N, Target, Node, and Target ratio columns.