Yuan Zhongshang, Ji Jiadong, Zhang Xiaoshuai, Xu Jing, Ma Daoxin, Xue Fuzhong
Department of Biostatistics, School of Public Health, Shandong University, Jinan 250012, China.
Department of hematology, Qilu hospital of Shandong University, Jinan 250012, China.
Sci Rep. 2016 Sep 30;6:34159. doi: 10.1038/srep34159.
Complex disease is largely determined by a number of biomolecules interwoven into networks, rather than a single biomolecule. Different physiological conditions such as cases and controls may manifest as different networks. Statistical comparison between biological networks can provide not only new insight into the disease mechanism but statistical guidance for drug development. However, the methods developed in previous studies are inadequate to capture the changes in both the nodes and edges, and often ignore the network structure. In this study, we present a powerful weighted statistical test for group differences of directed biological networks, which is independent of the network attributes and can capture the changes in both the nodes and edges, as well as simultaneously accounting for the network structure through putting more weights on the difference of nodes locating on relatively more important position. Simulation studies illustrate that this method had better performance than previous ones under various sample sizes and network structures. One application to GWAS of leprosy successfully identifies the specific gene interaction network contributing to leprosy. Another real data analysis significantly identifies a new biological network, which is related to acute myeloid leukemia. One potential network responsible for lung cancer has also been significantly detected. The source R code is available on our website.
复杂疾病很大程度上是由交织在网络中的多种生物分子决定的，而非单一生物分子。不同的生理状况，如病例组和对照组，可能表现为不同的网络。生物网络之间的统计比较不仅能为疾病机制提供新见解，还能为药物研发提供统计指导。然而，先前研究中开发的方法不足以捕捉节点和边的变化，且常常忽略网络结构。在本研究中，我们提出了一种针对有向生物网络组间差异的强大加权统计检验方法，该方法独立于网络属性，能够捕捉节点和边的变化，并且通过对位于相对更重要位置的节点差异赋予更大权重，同时考虑网络结构。模拟研究表明，在各种样本量和网络结构下，该方法比先前方法具有更好的性能。对麻风病全基因组关联研究（GWAS）的一项应用成功识别出了导致麻风病的特定基因相互作用网络。另一项真实数据分析显著识别出了一个与急性髓系白血病相关的新生物网络。一个可能与肺癌相关的网络也已被显著检测到。我们网站上提供了源R代码。
