基于支持向量回归算法的抗乳腺癌药物活性预测  

Prediction of Anti-breast Cancer Drug Activity Based on Support Vector Regression Algorithm

作　　者：黄甜 魏静雯 HUANG Tian;WEI Jingwen(School of Applied Mathematics,Nanjing University of Finance and Economics,Nanjing 210000,China)

机构地区：[1]南京财经大学应用数学学院,江苏南京210000

出　　处：《新乡学院学报》2022年第6期10-14,共5页Journal of Xinxiang University

基　　金：江苏省研究生科研创新计划项目(KYCX21_1499)。

摘　　要：根据候选药物中分子描述符对应的生物活性指数,利用条件互信息最大化准则选出了20个具有显著影响的特征分子描述符。借助所选特征分子描述符建立了支持向量回归算法,并对新合成化合物的生物活性进行预测和分析。利用sklearn中metric模块对预测结果进行了拟合,验证了预测结果的正确性。According to the biological activity of candidate drug molecular descriptors, 20 characteristic molecules with significant impact are selected by means of the conditional mutual information maximization criterion. The support vector regression algorithm is established with the help of the selected characteristic molecular descriptor, and the bioactivity of the newly synthesized compounds is predicted and analyzed. The prediction results are fitted by using the metric module in sklearn, and the correctness of the prediction results is verified.

关 键 词：支持向量回归 特征选择 条件互信息最大化 乳腺癌候选药物 

分 类 号：O242[理学—计算数学]