In order to accurately determine the directions of being constructed gene regulatory network, on the basis of path consistency algorithm based on conditional mutual information (PCA-CMI), the authors applied node ordering (NO) to design a novel algorithm, PCA-CMI-NO. To achieve this purpose, they made some improvements on the graph splitting method: first filtered mutual information between a pair of nodes, then arranged substructures in descending order of Bayesian scores, and finally according to the arrangement chose the orientation of the edge between the same gene pair included in the different substructures. In the end, the ordering results were employed in the regulatory network inferred by PCA-CMI to obtain the directions, and at the same time the conditional mutual information was used to remove the edges of independent relationship to improve the accuracy of the gene regulatory network. DREAM3 datasets were used to compare the performance of PCA-CMI-NO and the ordered K2 algorithm. The superiority of PCA-CMI-NO was proved.%为了在构建基因调控网络时能确定网络方向,在基于条件互信息的路径一致性算法PCA-CMI的基础上,利用节点拓扑排序(node ordering,NO)建立了构建调控网络的PCA-CMI-NO算法.为建立这一算法,对图分裂方法加以改进:首先对基因对间的互信息进行筛选,然后按贝叶斯得分对子图排序,根据子图顺序选取不同子图中含相同基因对间边的方向,从而确定基因表达数据中节点的顺序.最后,将节点拓扑排序结果应用于PCA-CMI所构建的网络,获得有向网络,同时,使用条件互信息去除独立关系的边,以提高网络准确率.采用DREAM3数据集,将PCA-CMI-NO算法与有序的K2算法进行对比,验证了算法的优越性.
展开▼
