Information content analysis reveals desirable aspects of in vivo experiments of a synthetic circuit
Information content analysis reveals desirable aspects of in vivo experiments of a synthetic circuit
复制标题
DOI：
10.1109/cibcb.2019.8791449
复制
发表时间：
2019-08
期刊：
影响因子：
--
通讯作者：
D. G. Cabeza;L. Bandiera;E. Balsa-Canto;F. Menolascina
中科院分区：
文献类型：
--
作者：
D. G. Cabeza;L. Bandiera;E. Balsa-Canto;F. Menolascina
合成生物学基于合理设计生物系统以丰富其新功能的范式。与其他工程学科概述的路径不同，合成生物学迄今为止对数学模型的使用有限。事实上，他们的费力推理利用了噪声数据，这些数据通常可以提供对系统行为的部分洞察。如果通过扰动生成的数据质量是推理的障碍，那么我们应该如何分析刺激的信息量？哪些因素促成了它？在这里，我们结合系统识别和贝叶斯推理的思想来量化体内实验的信息内容，以校准遗传切换开关的确定性模型。除了在贝叶斯和费舍尔基于信息的推理方法之间建立联系之外，我们发现 $\sim 40\%$ 增益可以归因于扰动方案的动态特性。我们的结果暗示了将贝叶斯实验设计纳入合成电路表征的重要性。
Synthetic biology is predicated upon the paradigm of rationally engineering biological systems to enrich them with new functions. Departing from the path outlined by other engineering disciplines, synthetic biology has made limited use of mathematical models so far. Indeed, their laborious inference leverages on noisy data that often provide partial insights into the system behaviour. If the quality of data generated via a perturbation is a road-block to inference, how should we analyse the informativeness of a stimulus? Which are the factors that contribute to it? Here we combine ideas from System Identification and Bayesian inference to quantify the information content of in vivo experiments for the calibration of a deterministic model of the genetic toggle switch. Beyond establishing a link between Bayes- and Fisher Information-based inference approaches, we find a $\sim 40\%$ gain can be ascribed to the dynamical properties of the perturbation scheme. Our results hint at the importance of including Bayesian experimental design in the characterisation of synthetic circuits.