Table 5

Comparison of different clustering algorithms

Analyzed data set
SPC
MCL
GSPC



clustered cases
error,1 %
clustered cases
error,1 %
clustered cases
error,1 %

SCOP domains
8666 (94%)2
9.3
9208
12.2
9276 (101%)
7.7
SwissProt InterPro domains
96716 (99%)
15.6
97792
14.3
103729 (106%)
10.7
SwissProt keywords
98276 (99%)
21.8
99636
20.8
105339 (106%)
15.7
Bacterial genomes, FunCat 1.3
4652 (103%)
14.1
4517
14.7
5043 (112%)
14.8

1-The error is defined as error=100%*(FP+FN)/(TP+FN). 2-In parentheses the percentage of clustered sequences relative to the corresponding numbers of the MCL algorithm (100%) are indicated.

Tetko et al. BMC Bioinformatics 2005 6:82   doi:10.1186/1471-2105-6-82