Orengo C A, Taylor W R
Department of Mathematical Biology, National Institute of Medical Research, Mill Hill, London, U.K.
J Theor Biol. 1990 Dec 21;147(4):517-51. doi: 10.1016/s0022-5193(05)80263-2.
A reduction in the time required to compare two protein structures has been achieved for a previously developed structure alignment method, by reducing the number of residue pair comparisons which must be performed between the two structures. Subsets of residue pairs are selected by an iterative procedure. Initially, selection is based on similarities in solvent accessible surface areas or torsional angles or a combination of both properties, giving subsets containing approximately 2% of the total number of residue pairs. Using these subsets, a rough comparison of the two structures is generated by the structural alignment program. The information returned from this can be used to identify more accurately topologically equivalent residues in the two proteins, thus enabling a new and much smaller subset (less than 0.2% of the total number of residue pairs) to be selected. The process of iterative refinement of the residue pair subsets is repeated once more, when in 95% of the structure comparisons tested, the correct alignment of the proteins was obtained. Times required to compare the structures using the refined subsets are insignificant compared to the initial comparison, so that considerable increases in speed are possible. The method was tested on two groups of proteins, a set of remotely related alpha/beta nucleotide proteins and the variable and constant domains of the immunoglobulins. Increases in speed ranging from 50-fold to greater than 150-fold were obtained depending on the degree of similarity of the two structures. In some comparisons the alignment was improved due to the reduction in noise obtained by comparing mainly equivalent residues.
对于先前开发的一种结构比对方法，通过减少两个结构之间必须进行的残基对比较数量，实现了比较两个蛋白质结构所需时间的缩短。残基对的子集通过迭代过程进行选择。最初，选择基于溶剂可及表面积或扭转角的相似性或这两种属性的组合，得到包含大约2%的残基对总数的子集。使用这些子集，结构比对程序对两个结构进行粗略比较。由此返回的信息可用于更准确地识别两种蛋白质中拓扑等效的残基，从而能够选择一个新的且小得多的子集（少于残基对总数的0.2%）。当在95%的测试结构比较中获得了蛋白质的正确比对时，残基对子集的迭代细化过程再重复一次。与初始比较相比，使用细化子集比较结构所需的时间微不足道，因此速度有可能大幅提高。该方法在两组蛋白质上进行了测试，一组是远缘相关的α/β核苷酸蛋白，另一组是免疫球蛋白的可变区和恒定区。根据两个结构的相似程度，速度提高了50倍至大于150倍。在一些比较中，由于主要比较等效残基而减少了噪声，比对得到了改善。
