Tableau-based protein substructure search using quadratic programming

BMC Bioinformatics

Table 7 Comparison of methods for substructure searching.

Fold	SCOP sid	P	S	T	Q	R	P/R	S/R	T/R	Q/R
β-grasp	d1ubia_	33	9	42	14	15	15/15	9/15	13/15	10/15
Key-barrel	d1tttb1	10	3	0	17	5	5/5	1/5	0/5	4/5
Immunoglobulin	d1ae6h1	27	1	4	1	11	9/11	1/11	4/11	1/11
Plait (ferredoxin)	d1bhne_	20	1	61	14	28	7/28	1/28	24/28	6/28
GFP-like	d1h6rb_	1	1	58	21	1	1/1	1/1	1/1	1/1
Jelly-roll	d2phlb1	1	1	19	15	12	1/12	1/12	10/12	5/12
TIM-barrel	d1tima_	16	16	40	33	32	16/32	16/32	28/32	32/32
NAD-binding fold	d1f6dc_	1	1	42	19	8	1/8	1/8	7/8	5/8

The number of superfamilies in the ASTRAL SCOP 95% sequence identity nonredundant database found for the eight query structures by ProSMoS (P), SSM (S), TOPS (T) and our method, QP tableau search (Q). ProSMoS and SSM return a list of significant hits, so the number of superfamilies found by these hits is shown. Other methods return scores for all database structures, and the number of superfamilies in the hits with a Z-score ≥ 3.0 are shown. The R column is the number of superfamilies, from the combined list of those found by all methods, in which the relevant query pattern or fold description is explicitly mentioned in the SCOP structure description. The ratio columns show the number of superfamilies in the R column found by the specified method.

ISSN: 1471-2105