Optimal identification of biochemical reaction networks
Optimal identification of biochemical reaction networks
复制标题
DOI：
10.1016/s0006-3495(04)74201-0
复制
发表时间：
2004-03-01
影响因子：
3.4
通讯作者：
Rabitz, H
中科院分区：
文献类型：
--
作者：
Feng, XJ;Rabitz, H
生物技术和计算机科学的进步为构建复杂生物网络的数学模型和系统地了解它们的性质提供了可能。然而，传统的网络辨识方法不能从有噪声的实验室测量中准确地恢复模型参数。本文介绍了最优辨识的概念，它利用一种全局反演算法来提取与实验室数据一致的参数的全部分布。此外，OI以闭环的方式将合适的计算算法与实验能力相结合，以最大限度地减小提取的参数分布的宽度。闭环OI程序寻找控制化学通量和数据观测的最佳集合，以主动过滤实验噪声并提高对所需参数的灵敏度。通过这种方式，可以通过对定制的实验室数据进行倒置来获得最高质量的网络参数。通过识别一个模拟的tRNA校对机制来说明OI的操作，在该机制中，与次优和非优方法相比，OI为所有的速率常数提供了更好的解。
Advances in biotechnology and computer science are providing the possibility to construct mathematical models for complex biological networks and systematically understand their properties. Traditional network identification approaches, however, cannot accurately recover the model parameters from the noisy laboratory measurements. This article introduces the concept of optimal identification (OI), which utilizes a global inversion algorithm to extract the full distribution of parameters consistent with the laboratory data. In addition, OI integrates suitable computational algorithms with experimental capabilities in a closed loop fashion to maximally reduce the breadth of the extracted parameter distribution. The closed loop OI procedure seeks out the optimal set of control chemical fluxes and data observations that actively filter out experimental noise and enhance the sensitivity to the desired parameters. In this fashion, the highest quality network parameters can be attained from inverting the tailored laboratory data. The operation of OI is illustrated by identifying a simulated tRNA proofreading mechanism, in which OI provides superior solutions for all the rate constants compared with suboptimal and nonoptimal methods.