Skip to main content

Table 1 Evaluation of protein clustering tools. The F-measure (between 0 and 1) measures the agreement between a clustering resulting from a given algorithm and a reference clustering provided with the dataset. An F-measure of 1 indicates perfect agreement. ASTRAL95_1_161 and ASTRAL95_2_161 refer to the two datasets of SCOP v1.61 used by Paccanaro et al. for spectral clustering [7]. All reported values, except for our algorithm FORCE and for Affinity Propagation, are from the same reference.

From: Large scale clustering of protein sequences with FORCE -A layout based heuristic for weighted cluster editing

Dataset

Method

F-measure

ASTRAL95_1_161

FORCE

0.85

ASTRAL95_1_161

Spectral clustering

0.81

ASTRAL95_1_161

Affinity Propagation

0.65

ASTRAL95_1_161

GeneRAGE

0.47

ASTRAL95_1_161

TribeMCL

0.32

ASTRAL95_1_161

Hierarchical clustering

0.26

ASTRAL95_2_161

FORCE

0.89

ASTRAL95_2_161

Spectral clustering

0.82

ASTRAL95_2_161

Affinity Propagation

0.69

ASTRAL95_2_161

GeneRAGE

0.54

ASTRAL95_2_161

TribeMCL

0.52

ASTRAL95_2_161

Hierarchical clustering

0.42