Improved protein structure reconstruction using secondary structures, contacts at higher distance thresholds, and non-contacts

Table 1 Comparison of the best of 20 models reconstructed using CONFOLD with the best of 20 models reconstructed using Reconstruct on the 12 benchmark proteins

PDB code - chain ID	SCOP class	L	Reconstruct			CONFOLD
PDB code - chain ID	SCOP class	L	TM-score	RMSD	GDT-TS	TM-score	RMSD	GDT-TS
1bkr-A	all-α	109	0.88	1.54	81.02	0.89	1.61	85.42
1odd-A	all-α	118	0.85	1.62	78.75	0.87	1.56	83.75
1cem-A	all-α	363	0.81	2.20	63.91	0.96	1.53	80.79
1pzc-A	all-β	123	0.91	1.38	85.04	0.91	1.28	84.84
1onl-A	all-β	128	0.91	1.42	83.86	0.91	1.39	84.65
1eur-A	all-β	365	0.83	2.04	68.98	0.96	1.42	83.38
1e6k-A	α/β	130	0.89	1.75	82.50	0.91	1.42	82.69
1o8w-A	α/β	146	0.90	1.65	79.72	0.91	1.50	82.52
1ede-A	α/β	310	0.95	1.61	82.26	0.96	1.40	82.58
1r9h-A	α + β	135	0.85	1.83	78.60	0.87	1.75	81.14
1ugm-A	α + β	125	0.85	1.88	77.21	0.87	1.71	80.53
1iu4-A	α + β	331	0.83	4.19	63.29	0.93	1.93	77.04
Average		199	0.87	1.93	77.10	0.91	1.54	82.44

Models are evaluated using TM-score, RMSD (in Å), and GDT-TS scores. Proteins are identified by their PDB ID followed by the chain ID. L is the length of the protein chain

ISSN: 1471-2105