Table 11.

Comparisons between the different methods on the Consolidated_3.19 network

		Method
		MCL	MCLO	MCL-CAw	CMC	HACO
	#Predicted	116	119	122	77	101
Wodak (#145)	#Matched	70	80	82	67	57
	Precision	0.603	0.672	0.672	0.870	0.564
	#Derived	79	80	82	67	64
	Recall	0.545	0.552	0.566	0.462	0.441
MIPS (#157)	#Matched	48	65	68	56	40
	Precision	0.414	0.546	0.557	0.727	0.396
	#Derived	63	65	68	56	57
	Recall	0.401	0.414	0.433	0.357	0.363
Aloy (#76)	#Matched	54	56	57	45	44
	Precision	0.466	0.471	0.467	0.584	0.436
	#Derived	55	56	57	45	45
	Recall	0.724	0.737	0.750	0.592	0.592

Methods considered: MCL, MCLO, MCL-CAw, CMC and HACO. CMC performed the best in terms of precision, while MCL-CAw performed the best in recall. #Matched: #Predictions matching some benchmark complex(es). #Derived: #Benchmark complexes derived by some predicted complex(es).

The Consolidated_3.19 network

#Proteins 1622; #Interactions 9704