一种基于网络表示学习的miRNA-疾病关联预测方法

miRNA-disease association prediction based on network representation learning method

针对miRNA-疾病关联研究中信息使用不充分、过于依赖网络中节点的相似度信息以及预测准确度较低的问题,提出一种基于网络表示学习的miRNA-疾病关联预测方法(network representation learning miRNA-disease association,NRLMDA)。该方法通过引入长链非编码RNA(lncRNA)构造出miRNA-lncRNA-疾病异构网络,丰富原有网络的生物学信息;采用网络表征学习node2vec算法在上述提出的异构网络中以一定的游走策略获得节点的近邻序列,并通过skip-gram模型进行深度学习,从而获得节点的低维特征向量;最后基于miRNA-miRNA相似性的关联规则推断方法预测miRNA与疾病的关联。该方法能够挖掘出全局网络的拓扑结构特征,并且不需要负样本。NRLMDA在留一交叉验证和五折交叉验证以及进一步的案例研究上的实验结果优于经典方法。

In view of the problem of inadequate use of information,excessive dependence on similarity information of nodes in the network and low prediction accuracy in miRNA-disease association studies,this paper proposed a miRNA-disease association prediction method based on network representation learning(NRLMDA:network representation learning miRNA-disease association).This method constructed a miRNA-lncRNA-disease heterogeneous network by introducing long-chain noncoding RNA(lncRNA),which enriched the biological information of the original network.It used the network representation learning node2vec algorithm in the heterogeneous network proposed above to obtain the node’s neighboring sequence with a certain walking strategy,and performed deep learning through the skip-gram model to obtain the low-dimensional feature vectors of the node.Finally,the association rule inference method based on miRNA-miRNA similarity predicted the association between miRNA and disease.This method could mine the topological structure characteristics of the global network without negative samples.NRLMDA’s experimental results on leave-one-out cross-validation and five-fold cross-validation as well as case stu-dies are superior to the classical methods.