Table 10.

Comparisons between the different methods on the FSW(Gavin+Krogan) network

		Method
		MCL	MCLO	MCL-CAw	CMC	HACO
	#Predicted	120	108	117	176	99
Wodak (#153)	#Matched	69	61	72	76	68
	Precision	0.575	0.564	0.615	0.432	0.687
	#Derived	78	72	83	84	77
	Recall	0.510	0.471	0.542	0.549	0.503
MIPS (#151)	#Matched	46	42	48	49	42
	Precision	0.383	0.388	0.410	0.278	0.424
	#Derived	61	55	66	65	56
	Recall	0.404	0.364	0.437	0.430	0.371
Aloy (#75)	#Matched	57	56	60	59	53
	Precision	0.475	0.518	0.513	0.335	0.535
	#Derived	57	56	60	57	53
	Recall	0.760	0.747	0.800	0.760	0.707

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

The FSW(Gavin+Krogan) network

#Proteins 1628; #Interactions 8688