Table 4.

Differences in each network-based methods.

Method	Dataset		Algorithm	AUC
LPBNI [31]	LPI	4870 lncRNA–protein interactions from NPInter database (2380 lncRNAs and 106 proteins)	Bipartite Network	0.8780
	PPI	×
	LLI	×
Yang et al. [33]	LPI	4883 lncRNA–protein interactions from NPInter database (1116 lncRNAs and 99 proteins)	A random walk model HeteSim	0.7972
	PPI	1608 protein–protein interactions from STRING database
	LLI	×
LPIHN [34]	LPI	10232 lncRNA–protein interactions from NPInter database (1113 lncRNAs and 99 proteins)	Random Walk with Restart	0.8839
	PPI	804 protein–protein interactions from STRING database
	LLI	lncRNA expression similarity from NONCODE 4.0 database (1113 lncRNA expression profiles)
Zheng et al. [32]	LPI	4467 lncRNA–protein interactions from NPInter database (1050 lncRNAs and 84 proteins)	SNF; A random walk model HeteSim	0.9068
	PPI	Sequence similarity from UniProt database; Functional annotation similarity from GO database; Protein domain similarity from Pfam database; STRING similarity from STRING database;
	LLI	×
PLPIHS [35]	LPI	lncRNA–protein interactions from GENCODE Release 24 (15941 lncRNAs and 20284 proteins) Co-expression data from COXPRESdb; Co-expression data from ArrayExpress and GEO; lncRNA–protein interactions from NPInter database;	SVM; A random walk model HeteSim	0.9678
	PPI	Protein–protein interactions from STRING database
	LLI	lncRNA co-expression similarity from NONCODE database (lncRNA expression profiles)

Bold representation performs best in AUC values and we found that the performance of the method is better when the heterogeneous network is composed by more sources. When heterogeneous networks are constructed by the same sources, the performance will be better for the heterogeneous networks constructed by weighted networks. ¹ https://github.com/USTC-HIlab/LPBNI (offline package); ² https://github.com/cyang235/LncADeep (offline package); ³ lncRNA–protein interactions; ⁴ protein–protein interactions; ⁵ lncRNA–lncRNA interactions; ⁶ A relevance search based on random walk in heterogeneous network to evaluate the relevance between a pair of lncRNA and protein, and a large relevance score means a high possibility that the lncRNA and protein interacts [94]. ⁷ Similarity Network Fusion: It is a nonlinear message-passing based method that iteratively updates each network and makes it more and more similar to the other [95].