Synthetic biology and machine learning for next generation biofuels

丙烷(C3H8)是一种挥发性碳氢化合物，具有非常有利的物理化学性质作为燃料，此外还有现有的全球市场和基础设施，用于广泛应用中的储存、分配和利用。因此，在旨在开发新的可再生替代品以补充目前使用的石油衍生燃料的研究中，丙烷是一个有吸引力的目标产品。该项目的重点是建造和评估用于生产可再生丙烷的替代微生物生物合成途径。这项研究将拓展可再生丙烷生产的代谢工具箱，为下一代生物燃料平台的发展提供新的见解和认识。该项目将专注于代谢工程的新生物催化部分。基于我们的ADO的晶体结构，我们已经确定了活性通道中的残基热点，当突变时会产生改进的变体(即更快的丙烷合成)。我们将通过构建ADO的合成DNA文库来组装酶库，在其中我们将增加整个酶的残基变化的频率。我们将使用曼彻斯特的内部SpeedyGenes和GeneGenie方法，这两种方法能够实现高保真的基因合成，并在整个蛋白质的残基上高效地生产错误校正的合成蛋白库，用于定向进化研究。重要的是，SpeedyGenes可以容纳多个(统计上)控制的组合变异序列，同时保持有效的酶纠错。我们将把这种制作合成文库的新方法与(I)用于主动学习序列-活性关系的机器学习方法，以及(Ii)我们在曼彻斯特已经/正在开发的HTP单细胞筛选方法相结合。该项目将基于MIB的新的BBSRC/EPSRC合成生物学中心(http://synbiochem.co.uk))，提供最先进的基础设施和合成生物学方法的培训。

Propane (C3H8) is a volatile hydrocarbon with highly favourable physicochemical properties as a fuel, in addition to existing global markets and infrastructure for storage, distribution and utilization in a wide range of applications. Consequently, propane is an attractive target product in research aimed at developing new renewable alternatives to complement currently used petroleum-derived fuels. This project focuses on the construction and evaluation of alternative microbial biosynthetic pathways for the production of renewable propane. This study will expand the metabolic toolbox for renewable propane production and provides new insight and understanding for the development of next-generation biofuel platforms. This project will focus on new biocatalytic parts for metabolic engineering. Based on our crystal structures of ADO we have already identified residue hotspots within the active channel that when mutated give rise to improved variants (i.e. faster propane synthesis). We will assemble enzyme libraries in which we increase the frequency of residue changes throughout the enzyme by constructing synthetic DNA libraries of ADO. We will use Manchester's in-house SpeedyGenes and GeneGenie methodologies, which enable high fidelity gene synthesis and efficient production of error-corrected synthetic protein libraries at residues throughout the protein for directed evolution studies. Importantly, SpeedyGenes can accommodate multiple and (statistically) controlled combinatorial variant sequences while maintaining efficient enzymatic error correction. We will couple this new approach of making synthetic libraries to (i) machine learning approaches for active learning of sequence-activity relationships, and (ii) HTP single cell screening approaches that we have/are developing at Manchester.The project will be based in the new BBSRC/EPSRC Synthetic Biology Centre in MIB (http://synbiochem.co.uk) providing state-of-the art infrastructure and training in synthetic biology methods.