Super paramagnetic clustering of protein sequences

BMC Bioinformatics

Table 1 Clustering of PDB sequences using SPC, gSPC and TRIBE-MCL algorithms

K	Cases	clusters	true positive	false positive	false negative	specificity	a posteriori sensitivity	a priori sensitivity
SPC
2	2472	479	2466	46	18	98.2	99.3	18.9
6	7332	1079	7107	276	274	96.3	96.3	54.4
20	8666	875	8324	413	401	95.3	95.4	63.7
all NN¹	8996	740	8507	586	548	93.6	93.9	65.1
TRIBE-MCL
	9208	964	8654	510	614	94.4	93.4	66.2
gSPC
6	7432	880	7252	277	239	96.3	96.8	55.5
20	8961	233	8709	377	314	95.9	96.5	66.6
all NN¹	9276	28	9009	392	329	95.8	96.5	68.9

1- the SPC analysis was performed using the complete similarity matrix and thus all Nearest Neighbors (NN) participated to the algorithm training.

ISSN: 1471-2105