kClust: fast and sensitive clustering of large protein sequence databases

BMC Bioinformatics

Table 2 Multidomain proteins clustering results

	#Clusters	#Corrupted clusters	%Corrupted clusters	#Wrong seq per	Time
				corrupted cluster
BLAST	6 977	186	2.66	1.5	3 h 21 m ^∗
kClust iter	8 537	10	0.1	1.1	13 m/39 m ^†
kClust sens	17 070	10	0.06	1.1	9 m
kClust 20	17 127	10	0.06	1.1	9 m
kClust 30	22 047	6	0.03	1.2	9 m
UCLUST 30	39 284	10132	25.79	1.7	30s
UCLUST 40	50 104	10512	20.98	1.4	40s
CD-HIT	29 163	6 234	21.37	1.89	43m

Clustering results on a set of 100 000 two-domain sequences constructed from 1000 domain architectures with 100 sequences each. ^∗The time is the sum of the run times of all 8 parallel threads. ^†Includes the time for the calculation of multiple sequence alignments.

ISSN: 1471-2105