Williamson M P, Kikuchi J, Asakura T
Krebs Institute, Department of Molecular Biology and Biotechnology, University of Sheffield, United Kingdom.
J Mol Biol. 1995 Apr 7;247(4):541-6. doi: 10.1006/jmbi.1995.0160.
We have developed a program that can calculate proton NMR chemical shifts for proteins, using a set of co-ordinates provided for example from an X-ray or NMR structure. When applied to NMR structures, agreement between calculated and observed shifts is generally of the same quality as that for crystal structures of resolution between 2.0 and 3.0 A. There is a rather weak correlation between standard deviation (SD) and the number of NMR constraints per residue, but none with the root-mean-square deviation of one NMR structure from another. Where minimised averaged structures are present, they have about the same SD as the population from which they were taken. Refinement methods such as energy minimisation and the use of relaxation matrices and back calculation produce little or no improvement in SD. The calculation has several applications, particularly as an independent means of measuring the quality of a structure (either in the crystal or in solution), and in identifying possible assignment errors.
我们开发了一个程序,该程序可以使用例如从X射线或核磁共振结构提供的一组坐标来计算蛋白质的质子核磁共振化学位移。当应用于核磁共振结构时,计算得到的位移与观测到的位移之间的一致性通常与分辨率在2.0至3.0埃之间的晶体结构相同。标准偏差(SD)与每个残基的核磁共振约束数量之间存在相当弱的相关性,但与一个核磁共振结构相对于另一个结构的均方根偏差无关。在存在最小化平均结构的情况下,它们的标准偏差与从中选取它们的总体大致相同。诸如能量最小化、使用弛豫矩阵和反演计算等优化方法对标准偏差几乎没有改善。该计算有多种应用,特别是作为一种独立的手段来衡量结构(无论是晶体结构还是溶液结构)的质量,以及识别可能的归属错误。