FunFOLD: an improved automated method for the prediction of ligand binding residues using 3D models of proteins

BMC Bioinformatics

Table 4 What is the effect on FunFOLD performance if different starting models are used?

Group ID	N	Mean score for group		Mean score for FunFOLD		Increase in mean score		Mean Z-score for group		Mean Z-score for FunFOLD		P-value (raw score)		P-value (Z-scores)		1 - p-value (raw score)		1 - p-value (Z-score)
		MCC	BDT	MCC	BDT	MCC	BDT	MCC	BDT	MCC	BDT	MCC	BDT	MCC	BDT	MCC	BDT	MCC	BDT
FN407	25	0.768	0.689	0.789	0.697	0.021	0.008	1.097	1.095	1.169	1.128	0.493	0.528	0.478	0.542	0.507	0.472	0.522	0.458
FN293	19	0.742	0.684	0.774	0.714	0.032	0.030	1.017	1.027	1.134	1.158	0.370	0.269	0.301	0.301	0.630	0.731	0.699	0.699
FN202	22	0.692	0.593	0.791	0.716	0.098	0.123	0.868	0.824	1.142	1.183	0.015	0.027	0.018	0.040	0.985	0.973	0.982	0.960

A repeat comparison of FunFOLD against the top 3 groups at CASP8, using TS1 3D models obtained from each comparison group. LEE_TS1 models were used for the FunFOLD comparison with group FN407, LEE-S_TS1 models for the comparison with FN293 and Phyre-de-novo_TS1 for the comparison with FN202.

ISSN: 1471-2105