Sotiropoulos Vassilios, Contou-Carrere Marie-Nathalie, Daoutidis Prodromos, Kaznessis Yiannis N
Department of Chemical Engineering and Materials Science, University of Minnesota, 151 Amundson Hall, 421 Washington Avenue S.E., Minneapolis, MN 55455, USA.
IEEE/ACM Trans Comput Biol Bioinform. 2009 Jul-Sep;6(3):470-82. doi: 10.1109/TCBB.2009.23.
Two very important characteristics of biological reaction networks need to be considered carefully when modeling these systems. First, models must account for the inherent probabilistic nature of systems far from the thermodynamic limit. Often, biological systems cannot be modeled with traditional continuous-deterministic models. Second, models must take into consideration the disparate spectrum of time scales observed in biological phenomena, such as slow transcription events and fast dimerization reactions. In the last decade, significant efforts have been expended on the development of stochastic chemical kinetics models to capture the dynamics of biomolecular systems, and on the development of robust multiscale algorithms, able to handle stiffness. In this paper, the focus is on the dynamics of reaction sets governed by stiff chemical Langevin equations, i.e., stiff stochastic differential equations. These are particularly challenging systems to model, requiring prohibitively small integration step sizes. We describe and illustrate the application of a semianalytical reduction framework for chemical Langevin equations that results in significant gains in computational cost.
在对生物反应网络系统进行建模时，需要仔细考虑其两个非常重要的特征。首先，模型必须考虑远离热力学极限的系统所固有的概率性质。通常，生物系统无法用传统的连续确定性模型进行建模。其次，模型必须考虑生物现象中观察到的不同时间尺度范围，例如缓慢的转录事件和快速的二聚化反应。在过去十年中，人们在开发用于捕捉生物分子系统动力学的随机化学动力学模型以及开发能够处理刚性的强大多尺度算法方面付出了巨大努力。在本文中，重点是由刚性化学朗之万方程（即刚性随机微分方程）控制的反应集的动力学。这些是特别具有挑战性的建模系统，需要极小的积分步长。我们描述并说明了一种用于化学朗之万方程的半解析简化框架的应用，该框架可显著降低计算成本。
