一种流形正则化非负矩阵分解的lncRNA-疾病关系预测方法  

LncRNA-disease Association Prediction Method Based on Manifold Regularized Non-negative Matrix Factorization

作　　者：董文文 林志毅[1] DONG Wen-wen;LIN Zhi-yi(School of Computer Science,Guangdong University of Technology,Guangzhou 510006,China)

机构地区：[1]广东工业大学计算机学院,广州510006

出　　处：《小型微型计算机系统》2022年第8期1589-1595,共7页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(618002072)资助;广东省自然科学基金项目(2018A030313389)资助;广州市科技计划项目(201902020012,201907010021)资助.

摘　　要：长链非编码RNA(lncRNA)在疾病的发生中起着重要作用,然而通过生物学实验探索lncRNA与疾病的关系昂贵且费时,必须开发出更准确和有效的计算方法来预测lncRNA与疾病的关系.本文提出了一种新的基于流形正则化非负矩阵分解的计算方法(MRNMFLDA)来预测lncRNA与疾病的关系.该方法首先采用相似性网络融合方法分别整合lncRNA与疾病的相似性,然后通过构建标签加权矩阵、引入流形正则化约束的非负矩阵分解算法来预测lncRNA与疾病潜在的关系.实验结果表明,本方法在留一交叉验证和5折交叉验证中AUC值分别达到0.8927和0.8635±0.0054,优于其他4种方法.案例研究表明,本方法能够有效地预测与3种疾病(肺癌,宫颈癌,和骨肉瘤)有关系的lncRNA.Long non-coding RNA(lncRNA)plays an important role in the occurrence of diseases.However,exploring the association between lncRNA and disease through biological experiments is expensive and time-consuming.It is necessary to develop more accurate and effective calculation methods to predict the association between lncRNA and disease.This paper proposes a new calculation method based on manifold regularized non-negative matrix factorization(MRNMFLDA)to predict the association between lncRNA and disease.First,the similarity network fusion method is used to integrate the similarities of lncRNA and diseases.Then,this paper constructs a label weighted matrix and introduces a manifold regularized non-negative matrix factorization algorithm to predict the potential association between lncRNA and disease.The experimental results show that the AUC value of this method in the leave-one-out cross-validation and 5-fold cross-validation achieve 0.8927 and 0.8635±0.0054,respectively,which outperform other four methods.Case studies show that this method can effectively predict the lncRNAs associated with three diseases(lung cancer,cervical cancer,and osteosarcoma).

关 键 词：lncRNA-疾病关系预测 非负矩阵分解 相似性网络融合 标签加权矩阵 流形正则化 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]