Towards the high-resolution protein structure prediction. Fast refinement of reduced models with all-atom force field

Table 2 The 7 protein subsets from the MOULDER testing set

PDB ID	SS type	L	RMSD range	Median RMSD	avg.ΔRMSD (best, rank.)	ΔRMSD
2mtaC	α	81	2.2–42.7	6.7	0.56 (0.30, 5)	0.26
1onc_	α, β	101	2.2–22.8	10.5	0.37 (0.25, 8)	0.30
1bbhA	α	127	2.5–20.8	6.5	0.51 (0.05, 17)	0.76
1mdc_	β	130	1.9–16.4	9.3	0.37 (0.13, 8)	0.02
1dxtB	α	143	2.0–34.1	7.2	0.56 (0.31, 3)	0.08
2fbjL	β	210	2.4–22.5	8.8	1.51 (0.32, 19)	0.53
2cmd_	α, β	310	2.5–20.2	5.8	1.26 (0.31, 13)	1.58

Particular columns contain: the PDB code, the secondary structure type, the protein length, the range of Cα RMSD (Å), the median of RMSD (Å), the average ΔRMSD of our method, in the brackets: the average ΔRMSD of the best method [12] and a ranking – the number of methods that outperformed our procedure (23 individual assessment methods were tested, SVMod that uses a composite score from the individual methods was not taken into account [12]), the ΔRMSD on the whole subset.

ISSN: 1472-6807