Table 4.

Summary of the appropriate usability and efficiency of all the methods by category when input and objectives are specified

Input and objectives	Method category											Suitable algorithms
	MSL	MUL	MSSL	MSD	PRSL	PRUL	PRSSL	PRSD	POSL	POUL	POSSL
Input: SOD/MOD having class labels of all samples. Objective: gene signature or marker/gene-signature-based classifier.	+	−	++	+	+	−	+	+	+	−	+	Machine-learning approach to integrate big data for precision medicine [70], Bimax biclustering [53], [54], CC biclustering [55] and spectral biclustering [59]
Input: SOD/MOD having all sample class labels. Objective: module/subnetwork detection.	+	++	++	×	×	++	+	+	+	++	+	iBAG [118], MCD [119], intNMF [72], iNMF [96], Joint NMF [91], [92], iCluster [94], iCluster+ [95], JIVE [137], ssCCA [99], Bimax biclustering [53], [54], CC biclustering [55], spectral biclustering [59], MDI [103], BCC [106] and SNF [109]
Input: SOD/MOD (big data) having some sample class labels but not all, or, SOD/MOD with all sample class labels but need for clustering toward both samples and genes together. Objective: gene classification signature module/subnetwork detection.	×	+	++	+	×	+	++	+	×	+	++	Bimax biclustering [53], [54], CC biclustering [55], XMotifs biclustering [58], spectral biclustering [59] and combinatorial gene marker discovery [79]
Input: SOD/MOD having all sample class labels. Objective: singular gene marker/hub gene/driver gene.	+	+	+	+	+	+	+	+	+	++	++	StatBicRM [68], epigenetic gene marker discovery through feature selection [41] and machine-learning approach to integrate big data for precision medicine [70]
Input: SOD/MOD with some sample class labels or SOD/MOD with all sample class labels but need for clustering toward both samples and genes together. Objective: singular gene marker/hub gene/driver gene.	×	+−	+−	+	×	+−	+−	+	×	+	++	StatBicRM [68]
Input: MOD with all sample class labels. Objective: Co-module of gene–drug.	×	++	×	+	×	×	×	+	×	×	×	SNPLS [102]
Input: MOD with all sample class labels. Objective: prognosis gene signature.	×	×	×	++	×	×	×	++	+	×	×	Net-Cox [50], netSVM [52], CoxPath [123], MKGI [124] and ATHENA [113]
Input: MOD with all sample class labels. Objective: Kernel-based classifier/regression model.	++	+	+	++	×	×	×	+	×	×	×	SDP/SVM [116,] FSMKL [117], penalized logistic regression model [51], sglasso [38], [40], fglasso [37], [39] and rMKL-LPP [111]
Input: MOD with all sample class labels. Objective: Pathway marker.	×	×	×	×	++	×	×	×	×	++	×	Pathway-based classification [71] and Significantly mutated pathway detection [82]
Input: SOD/MOD with all sample class labels. Objective: Feature/feature score.	+	+	+−	+−	+−	++	++	+	+−	++	+−	Anduril [120], CNAmet [23], Graph-based learning [122] and NBS [75].
Input: SOD/MOD with all sample class labels. Objective: closed frequent association rules or dense subgraphs or rule-based classifier.	+	+	++	++	×	×	++	×	+	+	++	iSubgraph [63], TrapRM [80], Lemon-Tree [110], ConGEMs [43], RiboFSM [60], StatBicRM [68] and normalized ImQCM [34], [35]
Input: MOD with all sample class labels. Objective: combinatorial gene marker.	×	×	×	×	×	×	++	×	×	×	×	Combinatorial gene marker discovery [79]
Input: MOD with all sample class labels. Objective: gene exclusive module.	×	×	×	×	×	×	×	×	×	×	++	MEMo [87]

‘++’, Best or highly useful; ‘+’, good or useful; ‘+−’, average, neutral or can be used; ‘−’, rarely used or poor; ‘×’, NA or cannot be used.