Luo Jiawei, Huang Wei, Cao Buwen
IEEE/ACM Trans Comput Biol Bioinform. 2018 Jan-Feb;15(1):309-315. doi: 10.1109/TCBB.2016.2612199. Epub 2016 Sep 21.
MicroRNAs (miRNAs) play an essential role in many biological processes by regulating the target genes, especially in the initiation and development of cancers. Therefore, the identification of the miRNA-mRNA regulatory modules is important for understanding the regulatory mechanisms. Most computational methods only used statistical correlations in predicting miRNA-mRNA modules, and neglected the fact there are causal relationships between miRNAs and their target genes. In this paper, we propose a novel approach called CALM(the causal regulatory modules) to identify the miRNA-mRNA regulatory modules through integrating the causal interactions and statistical correlations between the miRNAs and their target genes. Our algorithm largely consists of three steps: it first forms the causal regulatory relationships of miRNAs and genes from gene expression profiles and detects the miRNA clusters according to the GO function information of their target genes, then expands each miRNA cluster by greedy adding(discarding) the target genes to maximize the modularity score. To show the performance of our method, we apply CALM on four datasets including EMT, breast, ovarian, thyroid cancer and validate our results. The experiment results show that our method can not only outperform the compared method, but also achieve ideal overall performance in terms of the functional enrichment.
微小RNA（miRNA）通过调控靶基因在许多生物学过程中发挥着重要作用，尤其是在癌症的发生和发展过程中。因此，识别miRNA-信使核糖核酸（mRNA）调控模块对于理解调控机制至关重要。大多数计算方法在预测miRNA-mRNA模块时仅使用统计相关性，而忽略了miRNA与其靶基因之间存在因果关系这一事实。在本文中，我们提出了一种名为CALM（因果调控模块）的新方法，通过整合miRNA与其靶基因之间的因果相互作用和统计相关性来识别miRNA-mRNA调控模块。我们的算法主要包括三个步骤：首先从基因表达谱中形成miRNA与基因的因果调控关系，并根据其靶基因的基因本体（GO）功能信息检测miRNA簇，然后通过贪婪地添加（舍弃）靶基因来扩展每个miRNA簇，以最大化模块性得分。为了展示我们方法的性能，我们将CALM应用于包括上皮-间质转化（EMT）、乳腺癌、卵巢癌、甲状腺癌在内的四个数据集并验证我们的结果。实验结果表明，我们的方法不仅优于比较方法，而且在功能富集方面也能达到理想的整体性能。
