Taylor William R
Francis Crick Institute, 1 Midland Rd., London NW1 1AT, UK.
Comput Biol Chem. 2016 Oct;64:297-312. doi: 10.1016/j.compbiolchem.2016.06.007. Epub 2016 Aug 6.
An algorithm is described for the interaction of a hierarchy of objects that seeks to circumvent a fundamental problem in coarse-grained modelling which is the loss of fine detail when components become bundled together. A "currants-in-jelly" model is developed that provides a flexible approach in which the contribution of the soft high-level objects (jelly-like) are employed to protect the underlying atomic structure (currants), while still allowing them to interact. Idealised chains were used to establish the parameters to achieve this degree of interaction over a hierarchy spanning four levels and in a more realistic example, the distortion experienced by a protein domain structure during collision was measured and the parameters refined. This model of steric repulsion was then combined with sets of predicted distance constraints, derived from correlated mutation analysis. Firstly, an integral trans-membrane protein was modelled in which the packing of the seven helices was refined but without topological rearrangement. Secondly, an RNA structure was 'folded' under the predicted constraints, starting only from its 2-dimensional secondary structure prediction.
本文描述了一种用于对象层次结构交互的算法，该算法旨在解决粗粒度建模中的一个基本问题，即当组件捆绑在一起时精细细节的丢失。开发了一种“果冻中的葡萄干”模型，该模型提供了一种灵活的方法，其中利用软高级对象（果冻状）的贡献来保护底层原子结构（葡萄干），同时仍允许它们相互作用。使用理想化链来建立参数，以在跨越四个层次的层次结构中实现这种相互作用程度，并且在一个更现实的示例中，测量了蛋白质结构域结构在碰撞期间经历的变形并对参数进行了优化。然后将这种空间排斥模型与从相关突变分析得出的预测距离约束集相结合。首先，对一种完整的跨膜蛋白进行建模，其中七个螺旋的堆积得到了优化，但没有拓扑重排。其次，仅从其二维二级结构预测开始，在预测约束下“折叠”RNA结构。
