Integrating experimental and literature protein-protein interaction data for protein complex prediction

BMC Genomics

Table 10 Performance comparison with other protein complex prediction methods.

PPIN	Methods	#Complexes	P	R	F	Sn	PPV	Acc
Attr. PPIN I	Our method (BP,MF,CC)	231	0.589	0.36	0.447	0.521	0.541	0.531
	Our method (BP,MF)	182	0.659	0.326	0.436	0.471	0.571	0.518
	ClusterONE	199	0.568	0.331	0.418	0.468	0.609	0.534
	COACH	326	0.525	0.333	0.406	0.44	0.547	0.49
	CMC	120	0.608	0.218	0.321	0.371	0.606	0.474
Gavin PPIN	HUNTER	69	0.87	0.206	0.333	0.386	0.508	0.443
	MCL	103	0.718	0.245	0.366	0.53	0.489	0.509
	MCODE	70	0.739	0.154	0.255	0.283	0.519	0.384
Attr. PPIN II	Our method (BP,MF,CC)	247	0.636	0.382	0.477	0.525	0.576	0.551
	Our method (BP,MF)	206	0.679	0.348	0.46	0.477	0.578	0.525
	ClusterONE	464	0.375	0.431	0.401	0.523	0.655	0.585
	COACH	345	0.617	0.343	0.441	0.432	0.544	0.485
Krogan PPIN	CMC	111	0.748	0.235	0.358	0.381	0.589	0.474
	HUNTER	74	0.865	0.199	0.323	0.374	0.569	0.462
	MCL	309	0.291	0.245	0.266	0.57	0.396	0.475
	MCODE	72	0.75	0.159	0.263	0.27	0.552	0.386

#Complexes refers to the number of predicted complexes. F: F-score, P: precision, R: recall. The highest F-score of each approach is shown in bold.

ISSN: 1471-2164