从代谢时间序列数据中逐步推断可能的动态通量分布。
Stepwise inference of likely dynamic flux distributions from metabolic time series data.
机构信息
Department of Biomedical Engineering, Georgia Institute of Technology, Atlanta, GA, USA.
出版信息
Bioinformatics. 2017 Jul 15;33(14):2165-2172. doi: 10.1093/bioinformatics/btx126.
MOTIVATION
Most metabolic pathways contain more reactions than metabolites and therefore have a wide stoichiometric matrix that corresponds to infinitely many possible flux distributions that are perfectly compatible with the dynamics of the metabolites in a given dataset. This under-determinedness poses a challenge for the quantitative characterization of flux distributions from time series data and thus for the design of adequate, predictive models. Here we propose a method that reduces the degrees of freedom in a stepwise manner and leads to a dynamic flux distribution that is, in a statistical sense, likely to be close to the true distribution.
RESULTS
We applied the proposed method to the lignin biosynthesis pathway in switchgrass. The system consists of 16 metabolites and 23 enzymatic reactions. It has seven degrees of freedom and therefore admits a large space of dynamic flux distributions that all fit a set of metabolic time series data equally well. The proposed method reduces this space in a systematic and biologically reasonable manner and converges to a likely dynamic flux distribution in just a few iterations. The estimated solution and the true flux distribution, which is known in this case, show excellent agreement and thereby lend support to the method.
AVAILABILITY AND IMPLEMENTATION
The computational model was implemented in MATLAB (version R2014a, The MathWorks, Natick, MA). The source code is available at https://github.gatech.edu/VoitLab/Stepwise-Inference-of-Likely-Dynamic-Flux-Distributions and www.bst.bme.gatech.edu/research.php .
CONTACT
mojdeh@gatech.edu or eberhard.voit@bme.gatech.edu.
SUPPLEMENTARY INFORMATION
Supplementary data are available at Bioinformatics online.
动机
大多数代谢途径包含的反应多于代谢物，因此具有广泛的化学计量矩阵，对应于无限多的可能通量分布，这些分布与给定数据集中代谢物的动力学完全兼容。这种未确定度给从时间序列数据定量刻画通量分布以及设计适当的预测模型带来了挑战。在这里，我们提出了一种逐步降低自由度的方法，得到了一种在统计学意义上可能接近真实分布的动态通量分布。
结果
我们将所提出的方法应用于柳枝稷木质素生物合成途径。该系统包含 16 种代谢物和 23 种酶反应。它有七个自由度，因此允许存在一个很大的动态通量分布空间，所有这些分布都能很好地拟合一组代谢时间序列数据。所提出的方法以系统和合理的生物学方式缩小了这个空间，并在仅仅几次迭代后收敛到一个可能的动态通量分布。所估计的解与在这种情况下已知的真实通量分布之间具有极好的一致性，从而为该方法提供了支持。
可用性和实施
该计算模型是用 MATLAB（版本 R2014a，MathWorks，马萨诸塞州纳蒂克）实现的。源代码可在 https://github.gatech.edu/VoitLab/Stepwise-Inference-of-Likely-Dynamic-Flux-Distributions 和 www.bst.bme.gatech.edu/research.php 获得。
联系方式
mojdeh@gatech.edu 或 eberhard.voit@bme.gatech.edu。
补充信息
补充数据可在 Bioinformatics 在线获得。
Zotero 插件