基于机器学习的miRNA与疾病关系预测系统的设计与实现  

Design and Implementation of a Prediction System for MiRNA and DiseaseRelationship Based on Machine Learning

作　　者：夏岱婷 彭荧荧[1] Xia Daiting;Peng Yingying(Hunan University of Chinese Medicine,Changsha 410001,China)

机构地区：[1]湖南中医药大学,长沙410001

出　　处：《黑龙江科学》2023年第8期41-43,69,共4页Heilongjiang Science

基　　金：湖南省教育厅科学研究一般项目“基于机器学习的miRNA与疾病关联预测研究”(19C1389)。

摘　　要：为挖掘miRNA与疾病之间的关联关系,进一步了解疾病病理,实现疾病防治,提出了一种基于机器学习的miRNA-疾病关联关系模型。基于miRNA-疾病关联关系、miRNA功能相似性信息、疾病语义相似性信息展开研究,通过高斯核函数获取高斯核相互作用谱相似性矩阵并进行区分,利用矩阵补全算法对稀疏的已知相似性矩阵进行补全,综合高斯核相互作用谱相似性矩阵、补全后的相似性矩阵及已知的相似性矩阵,得到综合相似性矩阵,根据线性邻域重建综合相似性矩阵,利用标签传播算法得到预测结果。该模型在全局留一法实验中,性能优于其他3种模型(IMCMDA、QIMCMDA、MCLPMDA)。根据该模型设计并实现了一款可视化的、前后端分离的miRNA-疾病关联关系预测系统,前端主要采用Vue框架和ElementPlus组件库实现,后端主要采用SpringBoot和MyBatis实现。对本系统进行业务功能和性能测试,均达到预期。In order to explore the correlation of miRNA and diseases,further know disease pathology,and realize disease prevention and control,the study proposes the miRNA-disease correlation model based on machine learning.Based on miRNA-disease association,miRNA functional similarity information,and disease semantic similarity information,research is carried out.Gaussian interaction profile kernel similarity matrix is obtained through Gaussian kernel function,and sparse known similarity matrix is completed using Matrix Completion algorithm.The comprehensive similarity matrix is obtained by integrating the Gaussian interaction profile kernel similarity matrix,the completed similarity matrix and the known similarity matrix,and then the comprehensive similarity matrix is reconstructed according to the linear neighborhood.The prediction results are obtained using the Label Propagation algorithm.The model performs better than the other three models(IMCMDA,QIMCMDA,MCLPMDA)in the experiments of global leave-one-out experiment.At the same time,according to the prediction results of the model,a visual,front-end and back-end separated miRNA-disease association prediction system is designed and implemented.The front-end of the prediction system is mainly implemented by Vue framework and Element-Plus component library,and the back-end is mainly implemented by SpringBoot and MyBatis.After business function and performance test,it is found that this system meet the expectation.

关 键 词：机器学习 miRNA-疾病关联关系 预测系统 VueSpringBoot 

分 类 号：R363[医药卫生—病理学]