摘要
母系胚胎亮氨酸拉链激酶(maternal embryonic leucine zipper kinase,MELK)因其在多种癌症的发生发展中发挥关键作用,已成为一个备受关注的抗肿瘤药物靶点。为高效挖掘MELK抑制剂,结合机器学习方法筛选大规模化合物库,力求发现高活性候选分子。首先,基于已知的MELK抑制剂数据构建并优化机器学习模型,其中随机森林回归为最佳模型(R^(2)=0.8004,RMSE=0.4968)。随后基于该模型,从筛选出的预测得分最高的5个候选化合物出发,进一步开展分子对接及分子动力学模拟研究,结果显示这些化合物在活性口袋中具有稳定的构象和良好的结合自由能,显示出潜在的MELK抑制活性。最后利用SwissADME对化合物进行全面的ADME性质预测。该研究验证了机器学习辅助虚拟筛选方法在MELK抑制剂发现中的有效性,为新型MELK抑制剂的设计和开发提供了重要候选分子及理论基础。
Maternal embryonic leucine zipper kinase(MELK)has become a focus of anti-tumor drugs because it plays a key role in the occurrence and development of many cancers.In order to mine MELK inhibitors efficiently,this study combined with machine learning method to screen large-scale compound libraries and tried to find high-activity candidate molecules.Firstly,t he machine learning model was constructed and optimized based on the known MELK inhibitor data,i n which Random Forest Regression is the best model(R^(2)=0.8004,RMSE=0.4968).Then,based on this model,five candidate compounds were selected with the highest prediction scores,and further carried out molecular docking and molecular dynamics simulation research.The research shows that these compounds have stable conformation and good binding free energy in the active pocket,showing potential MELK inhibition activity.Finally,
作者
黄池
魏荣斌
马少杰
HUANG Chi;WEI Rongbin;MA Shaojie(School of Pharmacy,Jiangsu Ocean University,Lianyungang 222005,China;Jiangsu Key Laboratory of Marine Drug Active Molecular Screening,Jiangsu Ocean University,Lianyungang 222005,China)
出处
《江苏海洋大学学报(自然科学版)》
2026年第1期83-94,共12页
Journal of Jiangsu Ocean University:Natural Science Edition
基金
江苏海洋大学江苏省海洋药物活性分子筛选重点实验室开放基金资助项目(HY202302)。
