Combining specificity determining and conserved residues improves functional site prediction

BMC Bioinformatics

Table 7 Assessment of different methods.

Family	ET	Rate4site	FASS	MB	S
PF00108	C	C	-	-	no MW
PF00128	-	++C	n/a	n/a	n/a
PF00135	-	-	n/a	n/a	n/a
PF00215	-	++	no MW	-	no MW
PF00278	+	+C	-	-	no MW
PF00293	++C	+	n/a	n/a	n/a
PF00303	C	C	no MW	+	0
PF00348	++C	++C	-	++C	-
PF00351	+	+	n/a	n/a	n/a
PF00579	++	++	-	+C	no MW
PF00583	-	-	n/a	n/a	n/a
PF00590	++	++	no MW	-	-
PF00693	+C	+C	n/a	n/a	n/a
PF00755	++C	++C	n/a	n/a	n/a
PF00871	++C	++C	n/a	n/a	n/a
PF00896	+	-	n/a	n/a	n/a
PF00925	C	C	0	-	0
PF00962	++C	+C	no MW	++C	C
PF01014	++C	++C	no MW	++C	0
PF01048	+C	+C	-	-	-
PF01112	-	+	n/a	n/a	n/a
PF01135	++	++	n/a	n/a	n/a
PF01202	-	-	no MW	+	0
PF01227	+	-	-	-	0
PF01234	+C	+C	n/a	n/a	n/a
PF01293	+C	C	n/a	n/a	n/a
PF01467	-	-	0	-	0
PF01583	-	-	no MW	-	0
PF01656	++C	++C	no MW	-	no MW
PF01702	++	++	n/a	n/a	n/a
PF01712	++C	C	n/a	n/a	n/a
PF01747	++C	++C	++C	++C	-
PF02110	++C	+C	n/a	n/a	n/a
PF02223	++C	+	no MW	-	0
PF02274	++C	++C	-	C	-
PF02277	C	C	no MW	-	0
PF02353	++	++	n/a	n/a	n/a
PF02569	++C	C	n/a	n/a	n/a
PF02898	-	+	n/a	n/a	n/a
PF02901	++	++	n/a	n/a	n/a
PF03061	-	-	0	-	0
PF03171	-	-	no MW	-	0
PF03332	-	-	n/a	n/a	n/a
PF03414	++C	+C	n/a	n/a	n/a
Success rate	32/44	34/44	1/7	8/22	1/5

ET: evolutionary trace method [13, 14]; ConSurf: ConSurf method [9]; FASS, MB, S: three methods from the TreeDet package [32]. '++': SDPsite with both λ = 0.5 and λ = 1 performs significantly better (p-value < 0.01 in Mann-Whitney test), '+': SDPsite with either λ = 0.5 or λ = 1 performs significantly better, 'C': SDPsite with CPs only performs significantly better; 'n/a' method is not applicable, 'no MW': Mann-Whitney test is unreliable (less than 5 value in one of the sets), 0: method applicable, but no positions predicted, prediction considered not successful.

ISSN: 1471-2105