A density-based clustering approach for identifying overlapping protein complexes with functional preferences

BMC Bioinformatics

Table 5 Performance comparison of three versions of DCAFP

PPI network	Version of DCAFP	f -measure			Accuracy	MMR
PPI network	Version of DCAFP	Precision	Recall	f -measure	Accuracy	MMR
Krogan	Density only	0.77	0.29	0.42	0.49	0.16
	FPs only	0.62	0.22	0.32	0.41	0.1
	Both	0.54	0.46	0.5	0.51	0.23
Gavin	Density only	0.56	0.31	0.4	0.41	0.17
	FPs only	0.7	0.2	0.31	0.4	0.09
	Both	0.56	0.38	0.45	0.46	0.19
Collins	Density only	0.69	0.33	0.45	0.44	0.2
	FPs only	0.65	0.22	0.33	0.39	0.11
	Both	0.69	0.44	0.54	0.47	0.23
DIP Scere	Density only	0.52	0.31	0.39	0.39	0.17
	FPs only	0.27	0.62	0.38	0.39	0.2
	Both	0.39	0.6	0.47	0.46	0.27
DIP Hsapi	Density only	0.32	0.15	0.2	0.21	0.03
	FPs only	0.38	0.2	0.26	0.25	0.04
	Both	0.39	0.29	0.333	0.32	0.08

DCAFP with density only is the version of DCAFP that only considers the density property, DCAFP with functional preferences (FPs) only is the version of DCAFP that only considers the property of functional preferences, and DCAFP with both is the complete version of DCAFP.

ISSN: 1471-2105