最优化方法任分子识别中的应用

分子识别是指两个或多个分子靠非共价键专一地结合在一起．分子识别在物质的形成，细胞信号的传递，基因信息的表达和药物的设计等方面起重要的作用．我们首先对数学和计算方法在分子识别上的应用作了回顾，并从量子力学，经典分子力学和热力学上解释分子识别可转换成一类最优化问题．其次，我们指出了解决这类最优化问题的困难．最后，我们报告了在预报配体与蛋白质识别上所获得的一类新的选择方法．这类方法可将预报的成功率从63％提高到90％．

Optimization Applications in Molecular Recognition

Molecular recognition is the specific non-covalent binding of two or more molecules.Molecular recognition plays an important role in nature such as materials self-assembly,cellular signal transduction and the expression of genetic information,and drug design.Mathematical modeling and computational techniques from quantum mechanics to classical molecular mechanics are reviewed in understanding the molecular recognition.Mathematically the molecular recognition problem can be modeled as an optimization problem.Challenges in solving the optimization problem are reviewed.Novel clustering and binning selection schemes are reported to increase the flexible ligand-protein docking prediction success rate from 63％ to 90％.