Development of the protein-ligand docking program using multi-objective optimization and interactive optimization

本研究采用多目标优化方法进行蛋白质与配体对接计算。同时使用多个评分函数，选取Pareto最优解作为对接试验的解。此外，还使用了交互优化的概念，因为减少误报的数量是对接计算中最重要的问题之一。在交互优化中，利用实验数据和基于配体的方法获得的估计药效团等几个标准来消除假阳性。将多目标优化与交互优化相结合的方法应用于大型试验集，并应用于实际的药物设计试验。

In this study, multi-objective optimization was adopted for protein-ligand docking calculations. Multiple scoring functions were simultaneously used, and the Pareto optima were selected as the solutions of docking trials. In addition, the concept of the interactive optimization was also used, because the reduction of the number of the false positives is one of the most important problems in docking calculations. In interactive optimization, false positives were eliminated by using several criterion, such as experimental data and estimated pharmacophore obtained by ligand-based approach. The combination use of multi-objective optimization and interactive optimization was tested for large test set, and was applied for actual drug design trials.