miRNA与疾病关联关系预测算法
发布时间：2019-08-03 11:21
【摘要】：microRNAs(miRNAs)在生命进程中发挥着重要作用.近年来,预测miRNAs与疾病的关联关系成为一个研究热点.当前,计算方法整体上可以分为两大类:基于相似度度量的方法和基于机器学习的方法.前者通过度量网络中节点之间的关联强度预测miRNA-疾病关联,但需要构建高质量的生物网络模型;后者将机器学习相关算法应用到这个问题中,但需要构建高可信度的负例集合.基于以上困难和不足,提出了一种计算模型BNPDCMDA,用于预测miRNAs-疾病关联关系.该方法首先构建miRNA-疾病双层网络模型,然后利用miRNA的功能相似度对其进行基于密度的聚类,进而将二分网络投影应用于聚类后的miRNAs及疾病集合构成的miRNA-疾病双层子网中,最终完成对miRNA与疾病关联关系的预测.实验结果表明,采用留一交叉验证法得到的AUC值可达99.08%,明显优于当前其他高效方法.最后,采用BNPDCMDA方法对某些常见疾病所关联的miRNAs进行预测,实验结果获得了文献的支持,进一步表明了该方法的有效性.
[Abstract]:MicroRNAs (miRNAs) plays an important role in the process of life. In recent years, predicting the relationship between miRNAs and disease has become a hot research topic. At present, the calculation methods can be divided into two categories as a whole: the method based on similarity measure and the method based on machine learning. The former forecasts miRNA- disease association by measuring the correlation strength between nodes in the network, but it is necessary to construct a high quality biological network model, while the latter applies the machine learning correlation algorithm to this problem, but needs to construct a negative case set with high reliability. Based on the above difficulties and shortcomings, a computational model BNPDCMDA, is proposed to predict the association of miRNAs- diseases. In this method, the double-layer network model of miRNA- disease is constructed, and then the density-based clustering is carried out by using the functional similarity of miRNA, and then the binary network projection is applied to the double-layer subnet of miRNA- disease formed by clustering miRNAs and disease set, and finally the prediction of the relationship between miRNA and disease is completed. The experimental results show that the AUC value obtained by the method of residual cross verification can reach 99.08%, which is obviously better than that of other efficient methods at present. Finally, the BNPDCMDA method is used to predict the miRNAs associated with some common diseases, and the experimental results are supported by the literature, which further shows the effectiveness of the method.
【作者单位】： 北京建筑大学电气与信息工程学院;哈尔滨工业大学计算机科学与技术学院;
【基金】：国家自然科学基金(61571163,61532014,61671189,61402132) 国家重点基础研究发展计划(973)(2016YFC09019 02)~~
【分类号】：R3416;TP311.13
本文编号：2522524
[Abstract]:MicroRNAs (miRNAs) plays an important role in the process of life. In recent years, predicting the relationship between miRNAs and disease has become a hot research topic. At present, the calculation methods can be divided into two categories as a whole: the method based on similarity measure and the method based on machine learning. The former forecasts miRNA- disease association by measuring the correlation strength between nodes in the network, but it is necessary to construct a high quality biological network model, while the latter applies the machine learning correlation algorithm to this problem, but needs to construct a negative case set with high reliability. Based on the above difficulties and shortcomings, a computational model BNPDCMDA, is proposed to predict the association of miRNAs- diseases. In this method, the double-layer network model of miRNA- disease is constructed, and then the density-based clustering is carried out by using the functional similarity of miRNA, and then the binary network projection is applied to the double-layer subnet of miRNA- disease formed by clustering miRNAs and disease set, and finally the prediction of the relationship between miRNA and disease is completed. The experimental results show that the AUC value obtained by the method of residual cross verification can reach 99.08%, which is obviously better than that of other efficient methods at present. Finally, the BNPDCMDA method is used to predict the miRNAs associated with some common diseases, and the experimental results are supported by the literature, which further shows the effectiveness of the method.
【作者单位】： 北京建筑大学电气与信息工程学院;哈尔滨工业大学计算机科学与技术学院;
【基金】：国家自然科学基金(61571163,61532014,61671189,61402132) 国家重点基础研究发展计划(973)(2016YFC09019 02)~~
【分类号】：R3416;TP311.13
【相似文献】
相关期刊论文 前1条
1 杨叶猗;孙林;;MiRNA调节线粒体功能及其意义[J];国际病理科学与临床杂志;2011年03期
，本文编号：2522524
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2522524.html
