考虑基因表达影响的微生物制造过程鲁棒多目标运行优化

With the analysis of a lot of genomes and the development of advanced gene-editing techniques, synthetic microbial manufacture can produce a lot of difficultly-produced chemicals using the techniques of synthetic biology. The synthetic microbial manufacture has been a strategic development direction of China. Microbial manufacture is the typical batch process. For the operation optimization of microbial manufacture process, we should consider the gene expression regulation and optimization all the specifications based on the properties of the batch process. As a new research filed, only some exploratory results were reported for the optimization of microbial manufacture process, and there is no systematic and mature design method for this issue. Meanwhile, the referenced strategies for other batch processes are also limited. In this project, we focus on the microorganism manufacture process. With the consideration of the effect of gene-expression, we control the expression of the key gene and study the regulation of the genes of a cell to establish the gene-metabolism coupled metabolic model and to design the optimization operation condition of the microorganism manufacture based on the characteristics of batch process. This project aims to systematically study the structure analysis of the complex gene expression network, the model of metabolic process with complex information transfer, the uncertain multi-objective optimization of batch process. The goal of this project is to develop a method to design the dynamical regulation strategy for the metabolic processes of some model microorganisms. As a result, this project will provide an idea for the operation optimization of batch process.

随着大量基因组的解析及基因编辑技术的发展，利用合成生物学技术，微生物制造实现了许多难以合成的化合物的生产，成为国家战略发展方向。合成微生物制造是典型的间歇过程，它的运行优化应考虑微观的基因表达控制，结合间歇过程特点全面优化微生物制造指标。作为新兴研究领域，目前针对合成微生物制造的运行优化还只有一些探索性成果，缺乏系统性、成熟的设计方法，而可借鉴的间歇过程运行优化方案也不多。本项目以合成微生物制造过程为研究对象，考虑基因表达对细胞代谢影响，控制关键基因的表达，研究基因相互作用，建立基因-代谢耦联的微生物生长与产物合成模型，结合间歇过程特征进行运行优化设计，系统研究复杂基因表达调控网络的结构解析、微生物代谢过程建模以及不确定多目标优化问题求解的科学问题，以模式微生物发酵过程为例进行方法验证，提出一套针对合成微生物制造的多目标鲁棒运行优化方案设计方法，为间歇过程的运行优化提供一种技术思路。

本项目在国家自然科学基金下，针对合成生物学、基因大规模编辑背景下的微生物制造过程中存在建模、监测、控制优化问题，围绕强噪声数据下网络模型结构辨识、非线性过程监测、非线性过程参数辨识和多目标优化科学问题，主要进行了如下几个方面的研究。（1）研究了数据缺失环境下的基因网络建模，利用贝叶斯方法对基因缺失数据进行预测，然后利用XGBoost对基因结构进行推断，在基因辨识的金标准数据上进行了验证，证明了提出方法的有效性；（2）针对微生物制造过程的非线性问题，本项目研究了典型非线性过程的状态监测、和建模问题，提出了基于三阶无迹卡尔曼滤波方法的非线性过程监测方法和强噪声下基于变分贝叶斯的非线性过程参数估计方法，该方法解决了非线性概率传播的计算问题和非线性参数的估计问题；（3）针对微生物制造过程控制优化问题，研究了基于通量平衡方法的微生物制造过程建模方法，并针对宏观过程微生物制造的多目标优化问题，研究了基于信息熵的多目标优化问题，并在酿酒酵母产番茄红素的制造过程中进行了应用验证，结果表明提出的运行优化方法提高了产物产量、缩短了制造时间。在本项目的资助下，共发表论文SCI论文5篇，申请发明专利3项，培养硕士研究生7名，本科生10余名。