Method Development--Computer Vision Based Algorithms

为了能够进行数据库分析和对接，我们开发了在速度和能力上独一无二的新型、最先进的计算工具。这些包括不依赖序列顺序的多结构比对例程，它可以同时比较所有结构，并在工作站上用约一分钟的CPU时间找到最保守的结构基序；灵活的蛋白质结构比较的快速方法，无需预先定义铰链(几秒钟)；刚体和铰链弯曲对接算法，它考虑整个蛋白质表面，与其他方法相比只需几天时间；发现蛋白质表面空洞的算法的开发；药物数据库的比较，允许共价键的旋转。

To enable the database analysis and the docking, we develop novel, state of-the-art computational tools which are unique in speed and capabilities. These include a sequence order-independent multiple structure alignment routine that simultaneously compares all structures, and finds in around a minute of CPU on a workstation the optimally conserved structural motif; A rapid method for flexible protein structure comparisons with no predefinition of the hinges (in seconds); Rigid-body and hinge-bending docking algorithms which consider entire protein surfaces, and carry out docking in minutes as compared to days by other methods; Develoment of algorithms for finding cavities on protein surfaces; Comparisons of databases of drugs, allowing rotations on covalent bonds.