基于图神经网络和XGBoost的抗乳腺癌候选药物预测模型研究

Study on Anti-Breast Cancer Drug Candidate Forecast Model Based on Graph Neural Network and XGBoost

摘要: 药物研发过程中，通过筛选影响显著的化合物继而合成抗癌药物无疑能够保证抗癌药物研发的高效性和针对性。本文针对收集到的与乳腺癌相关的ER𝛂活性的1974种化合物，首先利用基于遗传算法的随机森林模型筛选出前20个对生物活性最具有显著影响的分子描述符，其次以此选择分子描述符变量构建定量预测模型得到预测结果，随后构建化合物的分类预测模型。结果表明该模型预测具有很强的实践意义，采用的预测策略是有效的，可为抗乳腺癌药物的研发提供借鉴。

Abstract: In the process of drug development, screening compounds with significant influence and then synthesizing anticancer drugs can undoubtedly ensure the high efficiency and pertinence of anticancer drug development. Aiming at the collected 1974 compounds with ERα activity related to breast cancer, the random forest model based on genetic algorithm was used to screen out the top 20 molecular descriptors with the most significant impact on biological activity, and then the molecular descriptor variables were selected based on this. A quantitative prediction model is constructed to obtain the prediction result, and then a classification prediction model of the compound is constructed. The results show that the model prediction has strong practical significance, and the prediction strategy adopted is effective, which can provide reference for the research and development of anti-breast cancer drugs.