抗乳腺癌候选药物的生物活性预测模型的构建
Construction of Bioactivity Prediction Models for Breast Cancer Candidate Drugs

乳腺癌是目前发病率较高的疾病之一，选择能够拮抗ERα活性的化合物对治疗乳腺癌具有重要的意义。本文综合考虑ERα拮抗剂的生物活性即pIC₅₀值(与生物活性具有正相关性)，筛选治疗乳腺癌的候选药物。为了构建能够使化合物对抑制ERα具有更好的生物活性的预测模型，本文通过随机森林算法和距离相关系数算法，寻求主要变量进行降维处理；通过建立基于BP神经网络的pIC₅₀预测模型并进行训练与验证，为寻找处理后变量的全局最优解，采用粒子群优化算法，以pIC₅₀的最大值作为目标函数，设定参数运行求得优化结果。研究结果表明，pIC₅₀的最大值与其对应的分子描述符都在合理的区间范围内，说明此次建立的模型具有一定的稳定性与合理性。
Breast cancer is one of the diseases with high incidence at present. It is important to select compounds that can antagonize ERα activity in the treatment of breast cancer. In this study, the bioactivity of ERα antagonists (pIC₅₀ value), which was positively correlated with biological activity, was considered to screen candidate drugs for breast cancer. In order to construct a compound that can inhibit ERα with better biological activity, this paper seeks the main variables for dimensionality reduction through random forest algorithm and distance correlation coefficient algorithm; Through the establishment, training and verification of the pIC₅₀ prediction model based on BP neural network, in order to find the global optimal solution of the processed variables, the particle swarm optimization algorithm is adopted, the maximum value of pIC₅₀ is taken as the objective function, and the parameters are set to run to obtain the optimization results. The results show that the maximum value of pIC₅₀ and its corresponding molecular descriptor are within a reasonable range, indicating that the model established this time has certain stability and rationality.