Table 2.

Properties of our seven data sets used for benchmark comparison of the algorithms

Data set	Number of classes	Average (SD) class size	Max, min class size	Number of sites	Site information	PDB ref	‘True’ sites
GPCR	77	26.8 (34)	189, 3	214	ligand	1GZM	T94, T97, E113, G114, A117, T118, G121, L125, C167, L172, F203, V204, M207, F208, H211, Y268, A269, A272, A292, F293, K296
GPCR-190	39	4.9 (3.8)	21, 2	like ‘GPCR’
LacI	15	3.6 (2.5)	12, 2	339	ligand and DNA	1EFA	T5, L6, S16, Y17, Q18, R22, N25, Q26, H29, Q54, A57, S61, L73, A75, P76, I79, N125, P127, D149, S191, S193, W220, N246, Q248, Y273, D274, T276, F293
Ras/Ral	2	44.5 (24.5)	69, 20	218	protein	5P21	I24, Q25, D30, E31, D33, I36, E37, Q43, L53, M67, Q70, D92
Rab5/Rab6	2	5.0 (1)	4, 6	163	protein	1R2Q	K42, G43, Q44, H46, E47, F48, Q49, E50, S51, H83, A86, M88, Y90, G92, A93, Q94, E117, L118, Q119, R120, Q121, A122, S123, P124, N125, I126, V127, K183
AQP/GLP	2	30.0 (18)	48, 12	430	protein	1FX8	L21, W48, V52, A65, H66, L67, V71, T137, Y138, P139, N140, P141, L159, I163, I187, G195, P196, L197, G199, F200, A201, M202
Smad	2	10.0 (2)	12, 8	211	protein	1KHX	L263, Q264, T267, Q284, Q294, P295, L297, T298, S308, E309, A323, V325, M327, I341, F346, P360, Q364, R365, Y366, W368, N381, R427, T430, S460, V461, R462, C463, M466

Data sets are the G-protein coupled receptors (GPCR) and a smaller version (GPCR-190), the LacI family of transcription factors, Ras super-family of small GTP-ases (Rab5 versus Rab6; Rab versus Ral), the aquaporins versus glycerol porins (AQP/GLP) and the Smad family of transcription factors [more details in (5,8)].