Max F. Perutz Laboratories, Department of Structural and Computational Biology, Centre for Molecular Biology, University of Vienna, Campus Vienna Biocenter 5, 1030 Vienna, Austria.
J Biomol NMR. 2012 Jun;53(2):149-59. doi: 10.1007/s10858-012-9632-7. Epub 2012 May 13.
A method for generating protein backbone models from backbone only NMR data is presented, which is based on molecular fragment replacement (MFR). In a first step, the PDB database is mined for homologous peptide fragments using experimental backbone-only data i.e. backbone chemical shifts (CS) and residual dipolar couplings (RDC). Second, this fragment library is refined against the experimental restraints. Finally, the fragments are assembled into a protein backbone fold using a rigid body docking algorithm using the RDCs as restraints. For improved performance, backbone nuclear Overhauser effects (NOEs) may be included at that stage. Compared to previous implementations of MFR-derived structure determination protocols this model-building algorithm offers improved stability and reliability. Furthermore, relative to CS-ROSETTA based methods, it provides faster performance and straightforward implementation with the option to easily include further types of restraints and additional energy terms.
提出了一种从仅含骨架的 NMR 数据生成蛋白质骨架模型的方法，该方法基于分子片段替换（MFR）。在第一步中，使用实验性仅含骨架数据（即骨架化学位移（CS）和残差偶极耦合（RDC））从 PDB 数据库中挖掘同源肽片段。其次，对该片段库进行实验约束的细化。最后，使用 RDC 作为约束，使用刚体对接算法将片段组装成蛋白质骨架折叠。为了提高性能，可以在此阶段包含骨架核 Overhauser 效应（NOE）。与以前的 MFR 衍生结构确定协议的实现相比，该建模算法提供了更高的稳定性和可靠性。此外，与基于 CS-ROSETTA 的方法相比，它提供了更快的性能和简单的实现，并可以轻松地包含其他类型的约束和附加能量项。
