Novel knowledge-based mean force potential at the profile level

BMC Bioinformatics

Table 5 The optimized results of probability threshold.

Probability threshold	Number of profiles	Distance_profile	Contact_profile	Dihedral_profile	Surface_profile
0.04	21355	-	0.828263	0.815943	0.899127
0.05	19868	-	0.826101	0.815291	0.900184
0.06	15935	-	0.825473	0.816991	0.900127
0.08	7444	-	0.825815	0.815693	0.898678
0.10	3145	-	0.827828	0.815039	0.900998
0.12	1442	-	0.826786	0.814627	0.899441
0.14	759	0.907069	0.82626	0.816084	0.899387
0.16	404	0.909359	0.826889	0.815488	0.899437
0.17	303	0.906705	0.82597	0.81585	0.899063
0.18	235	0.909907	0.824466	0.816644	0.899639
0.20	186	0.908468	0.828051	0.815918	0.899407
0.22	138	0.906744	0.823012	0.811962	0.896226
0.24	81	0.907125	0.825444	0.81052	0.895877
0.26	46	0.904767	0.823669	0.809967	0.896212
0.28	28	0.892552	0.816189	0.799421	0.887908
0.30	21	0.873879	0.782432	0.777818	0.867548
0.32	21	0.872684	0.781907	0.779257	0.866134

Given in the table are the average CP scores of profile-level statistical potentials at different probability threshold. The discrimination is performed on our structure models. The distance_profile, contact_profile, dihedral_profile and surface_profile refer to the profile-level statistical potentials of distance-dependent, contact, Φ/Ψ dihedral angle and accessible surface respectively. Note that for small P_hvalue (<0.12), the profile-level distance-dependent potentials cannot produce efficient output, because the parameters of this potential are proportional to the square of the number of profiles.

ISSN: 1471-2105