Park Chihyun, Yun So Jeong, Ryu Sung Jin, Lee Soyoung, Lee Young-Sam, Yoon Youngmi, Park Sang Chul
Well-Aging Research Center, Samsung Advanced Institute of Technology, Samsung Electronics, Suwon, South Korea.
Biomedical HPC Technology Research Center, Korean Institute of Science and Technology Information, Daejeon, South Korea.
BMC Syst Biol. 2017 Mar 15;11(1):36. doi: 10.1186/s12918-017-0417-1.
Cellular senescence irreversibly arrests growth of human diploid cells. In addition, recent studies have indicated that senescence is a multi-step evolving process related to important complex biological processes. Most studies analyzed only the genes and their functions representing each senescence phase without considering gene-level interactions and continuously perturbed genes. It is necessary to reveal the genotypic mechanism inferred by affected genes and their interaction underlying the senescence process.
We suggested a novel computational approach to identify an integrative network which profiles an underlying genotypic signature from time-series gene expression data. The relatively perturbed genes were selected for each time point based on the proposed scoring measure denominated as perturbation scores. Then, the selected genes were integrated with protein-protein interactions to construct time point specific network. From these constructed networks, the conserved edges across time point were extracted for the common network and statistical test was performed to demonstrate that the network could explain the phenotypic alteration. As a result, it was confirmed that the difference of average perturbation scores of common networks at both two time points could explain the phenotypic alteration. We also performed functional enrichment on the common network and identified high association with phenotypic alteration. Remarkably, we observed that the identified cell cycle specific common network played an important role in replicative senescence as a key regulator.
Heretofore, the network analysis from time series gene expression data has been focused on what topological structure was changed over time point. Conversely, we focused on the conserved structure but its context was changed in course of time and showed it was available to explain the phenotypic changes. We expect that the proposed method will help to elucidate the biological mechanism unrevealed by the existing approaches.
细胞衰老会不可逆地阻止人类二倍体细胞的生长。此外，最近的研究表明，衰老是一个与重要复杂生物过程相关的多步骤演变过程。大多数研究仅分析了代表每个衰老阶段的基因及其功能，而没有考虑基因水平的相互作用和持续受干扰的基因。有必要揭示衰老过程中受影响基因及其相互作用所推断的基因型机制。
我们提出了一种新颖的计算方法来识别一个整合网络，该网络从时间序列基因表达数据中描绘出潜在的基因型特征。基于所提出的称为扰动分数的评分度量，为每个时间点选择相对受干扰的基因。然后，将选定的基因与蛋白质 - 蛋白质相互作用整合，构建特定时间点的网络。从这些构建的网络中，提取跨时间点的保守边以形成共同网络，并进行统计检验以证明该网络可以解释表型改变。结果证实，两个时间点的共同网络平均扰动分数的差异可以解释表型改变。我们还对共同网络进行了功能富集，并确定其与表型改变高度相关。值得注意的是，我们观察到所识别的细胞周期特异性共同网络作为关键调节因子在复制性衰老中起重要作用。
迄今为止，从时间序列基因表达数据进行的网络分析一直集中在随着时间点拓扑结构如何变化。相反，我们关注的是保守结构，但其背景会随时间变化，并表明它可用于解释表型变化。我们期望所提出的方法将有助于阐明现有方法未揭示的生物学机制。
