How accurate and statistically robust are catalytic site predictions based on closeness centrality?

BMC Bioinformatics

Table 3 Dataset used in comparison of ligated and unligated pairs

Enzyme¹	Ligated vs. unligated²	Protein length	SCOP class	Correl³	RMSD (Å)⁴
4-oxalocrotonate tautomerase (1BJP, 4OTB)	1, 2, 0	59	α + β	0.985	0.4
Ribonuclease A (1RBN, 1RSM)	0, 2, 2	124	α + β	0.984	0.6
Xylanase II (1BVV, 1XNB)	1, 1, 0	185	β	0.997	0.2
Trpysin (1A0J, 1UTK)	1, 2, 0	245	β	0.983	1.1
Aminopeptidase (1IGB, 1AMP)	1, 0, 0	291	α/β	0.992	0.6
Phospholipase C (1AOD, 2PLC)	3, 0, 2	294	α/β	0.992	0.2
Deacetoxycephalosporin C synthase (1W2N, 1W28)	0, 1, 0	298	β	0.980	0.5
Chorismate mutase (3CSM, 2CSM)	2, 2, 0	300	α	0.955	1.9
Alginate lyase A1-III (1HV6, 1QAZ)	2, 0, 1	354	α	0.995	0.1
tRNA-guanine transglycosylase (1R5Y, 1PUD)	1, 0, 0	382	α/β	0.814	0.5
Nitric oxide synthase oxygenase (1M9R, 3NOS)	2, 2, 0	480	α + β	0.992	0.2
Luciferase (1BA3, 1LCI)	3, 1, 0	544	Multi.	0.948	0.5
Class I alpha-1;2-mannosidase (1G6I, 1DL2)	1, 3, 0	549	α	0.996	0.1
Average ³				0.970	0.5
Standard deviation				0.049	0.5
Correlation (Correl vs. RMSD)					-0.17

¹ SCOP protein name for each pair examined. The ligated and unligated PDB id's, respectively, are provided in parentheses. ² The three values (ligated, unligated, tie) tabulate the number of catalytic residues with higher rank ordering between the structural pair. ³ Linear correlation coefficient comparing the CC scores between each structural pair. ⁴ Pairwise RMSD comparing structure similarity within each pair. Surprisingly, there is no significant correlation between the last two columns.

ISSN: 1471-2105