基于蛋白质相互作用网络拓扑参数预测乳腺癌相关基因  

Prediction of breast cancer-related genes based on the topological features of proteinprotein interaction network

作　　者：周漩 李占潮 ZHOU Xuan;LI Zhanchao(School of Pharmacy Guangdong Pharmaceutical University Guangzhou 510006 China)

机构地区：[1]广东药科大学药学院,广东广州510006

出　　处：《广东药科大学学报》2018年第3期360-364,共5页Journal of Guangdong Pharmaceutical University

基　　金：国家自然科学基金资助项目(21675035)

摘　　要：目的建立预测乳腺癌相关基因的新方法,为乳腺癌发病机制及治疗靶点的研究提供理论基础。方法以基因在蛋白质相互作用网络中的拓扑参数为输入参数,支持向量机建模,预测乳腺癌相关基因,并进行生物功能富集分析。结果采用10-折交叉验证评价模型,马氏相关系数和预测精度分别为0.800 9和0.895 1,模型预测精度良好。富集分析结果表明乳腺癌相关基因与某些生物过程与分子功能高度相关。结论本文建立的新方法可有效预测乳腺癌相关基因,为其他疾病相关基因的预测提供了新的手段。Objective To develop a new computational method for predicting breast cancer-related genes and provide a basis for the study of pathogenesis and therapeutic targets of breast cancer. Methods The method was developed by support vector machine with the protein-protein interaction network topological features as input parameter to characterize genes for predicting candidate breast cancer-related genes.Results The performance of the support vector machine model was verified based on the 10-fold crossvalidation test and achieved appreciable results,in which the matthews correlation coefficient and accuracy were 0.800 9 and 0. 895 1, respectively. By enrichment analysis it was found that the identified breast cancer-related genes were highly related to some biological process and molecular function,which provided new clues for researching pathogenesis of breast cancer. Conclusion The proposed method may become a useful tool for identifying breast cancer-related genes, and also provide a reference for predicting other disease-related genes.

关 键 词：乳腺癌 蛋白质相互作用网络 拓扑参数 支持向量机 

分 类 号：R737.9[医药卫生—肿瘤]