Multi-modal protein conformational space optimization method based on composite structure features

The invention discloses a multi-modal protein conformational space optimization method based on composite structure features. The method comprises the following steps that: on the basis of an evolutionary algorithm framework, RosettaScore3 is taken as an optimized objective function, and statistics is carried out to obtain three types of structure features, including the individual distance spectrum, the individual secondary structure spectrum and the individual dihedral angle spectrum of a population individual. Through a multi-modal strategy, algorithm sampling diversity is improved, offspring individuals are selected according to the structure features, the defect that an energy model is inaccurate can be effectively eliminated, and a prediction structure with high accuracy can be obtained through iterative evolution. The method has the advantages of being high in sampling efficiency, low in complexity and high in prediction accuracy. 种基于复合结构特征的多模态蛋白质构象空间优化方法，包括以下步骤：基于进化算法框架，以RosettaScore3为优化目标函数，统计得到种群个体的