Molecular complex detection in protein interaction networks through reinforcement learning

BMC Bioinformatics

Table 2 The RL algorithm yields competitive accuracy compared to other community detection algorithms on hu.MAP 1.0

	FMM Precision	FMM Recall	FMM F-score	CMMF	UnSPA	Qi et al. F1 score	F-Grand K-Clique	F-weighted K-Clique
RL Algorithm	0.612	0.482	0.547	0.654	0.772	0.559	0.789	0.988
Super.Complex	0.835	0.457	0.591	0.720	0.803	0.658	0.991	0.999
ClusterONE + MCL	0.471	0.686	0.579	0.797	0.911	0.794	0.77	0.967
PC2P	0.535	0.589	0.562	0.625	0.571	0.418	0.627	0.891
MP-AHSA	0.421	0.498	0.459	0.424	0.513	0.397	0.516	0.823
DPCMNE	0.457	0.048	0.086	0.408	0.454	0.132	0.609	0.608

The learned complexes on hu.MAP 1.0 are evaluated against all the known cleaned CORUM complexes
FMM, F-similarity-based Maximal Matching; CMMF, Community-wise Maximum F-similarity-based F-score; UnSPA, Unbiased Sn-PPV Accuracy; SPA, Sn-PPV Accuracy

ISSN: 1471-2105