Low-resolution refinement tools in REFMAC5

Robert A Nicholls et al. Acta Crystallogr D Biol Crystallogr. 2012 Apr.

Abstract

Two aspects of low-resolution macromolecular crystal structure analysis are considered: (i) the use of reference structures and structural units for provision of structural prior information and (ii) map sharpening in the presence of noise and the effects of Fourier series termination. The generation of interatomic distance restraints by ProSMART and their subsequent application in REFMAC5 is described. It is shown that the use of such external structural information can enhance the reliability of derived atomic models and stabilize refinement. The problem of map sharpening is considered as an inverse deblurring problem and is solved using Tikhonov regularizers. It is demonstrated that this type of map sharpening can automatically produce a map with more structural features whilst maintaining connectivity. Tests show that both of these directions are promising, although more work needs to be performed in order to further exploit structural information and to address the problem of reliable electron-density calculation.

PubMed Disclaimer

Figures

Figure 1

Histograms of the interatomic distances in the structure with PDB code

2jhp

(Sutton et al., 2007 ▶) corresponding to (a) main-chain atoms only and (b) both main-chain and side-chain atoms. Distances corresponding to atom pairs separated by one chemical bond are shown in red, those separated by two bonds are shown in blue and all other atom pairs are shown in black.

Figure 2

Distance dependence of the distribution of interatomic distances (for main-chain atoms only) for the target structure

2jha

(Sutton et al., 2007 ▶) using the sequence-identical

2jhp

as the external reference. The image shows the interatomic distance in

2jha

plotted against the corresponding distance in

2jhp

. Distances corresponding to atom pairs separated by one chemical bond are shown in red, those separated by two bonds are shown in blue and all others are shown in black.

Figure 3

Depictions of superposed models

2jha

and

2jhp

(a) coloured blue (

2jha

) and green (

2jhp

) and (b) coloured according to structural conservation (r.m.s.d.) of side-chain atoms relative to the residues’ local coordinate frames. Residues with different side-chain conformations are coloured red (d > 1 Å), fading through orange (d ≃ 0.5 Å) to yellow (d ≃ 0 Å) for highly conserved side chains.

Figure 4

Representations of the main chains of various superposed models coloured by side-chain conformational similarity. Images correspond to a comparison of the model of

2jha

after refinement with main-chain and side-chain external restraints from

2jhp

(with w ext = 7.6, κ = 1.0 and r max = 4.2) and (a) the original model

2jha

, (b) the reference model

2jhp

and (c) the model refined without external restraints. Each image displays the globally superposed compared models, with residues coloured according to structural conservation (r.m.s.d.) of side-chain atoms relative to the residues’ local coordinate frames. Residues with different side-chain conformations are coloured red (_d_> 1 Å), fading through orange (d ≃ 0.5 Å) to yellow (d ≃ 0 Å) for highly conserved side chains.

Figure 5

R factor (solid lines) and R free (dashed lines) after 30 _REFMAC_5 refinement iterations starting from the model

2jha

plotted against (a) the external restraints weight w ext, (b) the Geman–McClure weight κ on a logarithmic scale and (c) the maximum external restraint length r max. Lines correspond to the original model (blue), the model refined without external restraints (red) and the model refined with external restraints from the reference structure

2jhp

(black), generating restraints for both main-chain and side-chain atoms. In each graph, the two parameters not being considered were fixed to the values that globally minimized R free, i.e. w ext = 7.6, κ = 1.0, r max = 4.2.

Figure 6

R factor (solid lines) and R free (dashed lines) after 40 _REFMAC_5 refinement iterations starting from the model

1ydz

plotted against (a) the external restraints weight w ext, (b) the Geman–McClure weight κ on a logarithmic scale and (c) the maximum external restraint length r max. Lines correspond to the original model (blue), the model refined with local NCS restraints but without external restraints (red) and the model refined with local NCS restraints and α-helical restraints (green), external main-chain restraints from

2w72

(grey) and external main-chain and side-chain restraints from

2w72

(black). In each graph, the two parameters not being considered were fixed to the values that globally minimized R free, i.e. w ext, κ and r max are 8.9, 13 and 4.3, respectively, for α-helical restraints, 19, 5.9 and 4.4, respectively, for main-chain restraints and 4.2, 0.94 and 4.4, respectively, for main-chain and side-chain restraints.

Figure 7

Visual effects of map sharpening on electron density. This example was taken from PDB entry

2r6c

. Images show the map with (a) no map sharpening, (b) map sharpening using the inverse filter (no regularization) and (c) a regularized sharpened map using the L 2-type Tikhonov regularizer, with sharpening coefficients integrated over B and α, as described in the text. The backbone trace of

2r6c

chain C is shown in green. The homologous structure

2r6a

chain A is shown in blue, superposed using residues formula image – from

2r6a

chain A. The image shows unmodelled density in

2r6c

that corresponds to a helix present in

2r6a

. Both sharpened maps show more features than the unsharpened map, with the regularized map giving more connectivity. Images were generated using CCP_4_mg (McNicholas et al., 2011 ▶).

References

1. Adams, P. D. et al. (2010). Acta Cryst. D66, 213–221. - PubMed
1. Afonine, P. V., Grosse-Kunstleve, R. W. & Adams, P. D. (2005). Acta Cryst. D61, 850–855. - PMC - PubMed
1. Bailey, S., Eliason, W. & Steitz, T. (2007). Science, 318, 459–463. - PubMed
1. Bentley, J. L. (1975). A Survey of Techniques for Fixed Radius Near Neighbor Searching. Technical Report SLAC-186 and STAN-CS-75-513. Stanford Linear Accelerator Center.
1. Berman, H. M. et al. (2002). Acta Cryst. D58, 899–907. - PubMed

Low-resolution refinement tools in REFMAC5 - PubMed (original) (raw)