MRI: Development of a fully automated, 1,000-MicroChemostat microfluidic system for parallel, independent, long-duration, machine-guided experiments

范德比尔特大学被授予开发Genesis的奖项，这是一个包含1,000个微恒化器的全自动微流控系统，用于平行、独立、长时间、机器指导的实验，以研究微生物细胞。微生物，如面包酵母和酿酒酵母，预计将在疫苗、癌症治疗和其他药物、食品蛋白质和化学工业原料的生产中发挥越来越大的作用，以及从大气中隔离二氧化碳，所有这些领域都是迫切的社会需求。Genesis公司将利用机器学习和人工智能(AI)作为一名“机器人科学家”或“自动驾驶实验室”，以加速开发描述微生物新陈代谢和生长的数学模型。这些将帮助我们理解，并可能优化，例如，组成人类微生物群的许多不同微生物物种之间的相互作用，这些微生物对健康和疾病都有贡献。Genesis自身创建细胞信号和新陈代谢计算模型的能力应该会促进医学、生物技术和基础生物学知识的发展，因为这些模型是优化实验和解释数据以揭示支配生物过程的规则所必需的。Genesis项目将涉及范德比尔特大学的三个既定研究和培训项目，所有这些项目都积极招募和参与本科生、研究生和博士后实习生参与其研究项目。Genesis提供了一系列非常有吸引力的技术挑战以及科学和社会机会，非常适合人工智能、群体机器人、机器学习、探索微生物和悬浮哺乳动物细胞中的信号和代谢途径的跨学科研究培训，以及解决如何实际上利用生物学的复杂性来解决营养、健康和医学方面的社会问题。Genesis将允许科学家和工程师及其受训人员通过增进我们对生物学和疾病的理解以及提高生物化学品和药品的工业生产效率来解决一些重要的科学、商业和社会问题。作为课堂教学的工具，Genesis还提供了额外的影响，因为它最终将允许学生提出问题，并要求Genesis设计和进行回答这些问题所需的实验。Genesis将被设计成以低成本大规模生产，这样小型实验室就可以负担得起一个小规模的系统。通常使用的是酵母分批工作，在那里它生长和繁殖，直到它耗尽食物或创造一个它不能再茁壮成长的环境。在研究实验室中培养一小批酵母，可能需要在多孔培养皿中的一口井中放置一毫升生长介质，而制药公司的酵母生物反应器可以容纳几千升，啤酒厂的酵母生物反应器可以容纳一百万升。作为间歇生物反应器的替代，被称为恒化器的连续流动生物反应器提供稳定的食物供应，并不断地去除多余的酵母甚至悬浮的哺乳动物细胞及其代谢物，以保持稳定的生长。人们越来越认识到，化学药物可以提供可重复性、可靠性和生物同质性的数据集，这些数据集非常适合探测活细胞的新陈代谢和信号。然而，将“自动驾驶”和机器学习技术应用于推进生物学知识，将受益于一千或更多个在计算机控制下并行运行的恒化器。无论是商业生产还是研究中的恒化器，都没有在大小、成本和自动化仪器方面的正确组合。Genesis系统将满足这一需求，方法是使用最先进的多通道微流体泵和阀来控制各种条件下的所有微型恒化器，并使用不同菌株的酵母、蜂群机器人一次移动48个微恒化器以及非常高吞吐量的质谱仪每10或15秒进行一次广泛的新陈代谢测量，这将产生超出人类控制、处理和解释能力的TB级数据。由此产生的计算模型可能有数千个方程式。Genesis将首次提供一种有效的手段来设计和进行大量的生物实验，这些实验需要对这些模型进行参数化、验证和利用，以探索甚至控制特定应用的生物系统。一旦Genesis投入使用，将立即开展的项目包括细胞信号和新陈代谢的基础研究，对共同培养的细菌物种的代谢组学相互作用的定量探索，这些细菌物种共同生产可用于食品和化工原料的蛋白质，跟踪多个平行的进化历史，以确定哪些环境和遗传因素对微生物合作的进化至关重要，以及改进使用哺乳动物细胞生产治疗性抗体的方法。两个Genesis仪器的并行开发，一个由NSF资助，另一个由瑞典哥德堡的Chalmers技术大学资助，将Vanderbilt在微流体和质谱学方面的专业知识与Chalmers在人工智能、机器学习和酵母方面的专业知识相结合，创造出一对机器人科学家，他们将加快对生活规则的调查，并发现一些社会紧迫问题的新解决方案。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

An award is made to Vanderbilt University to develop Genesis, a fully automated microfluidic system containing 1,000 microchemostats for parallel, independent, long-duration, machine-guided experiments to study microbial cells. Microbes such as the baker’s and brewer’s yeast Saccharomyces cerevisiae are expected to play an ever-increasing role in the production of vaccines, cancer therapies and other pharmaceuticals, food protein, and feedstock for the chemical industry, and sequestration of carbon dioxide from the atmosphere, all areas of pressing societal need. Genesis will use machine learning and artificial intelligence (AI) to serve as a “robot scientist” or “self-driving laboratory” to accelerate the development of mathematical models that describe microbial metabolism and growth. These will help us understand, and possibly optimize, for example, the interactions between the many different microbial species that comprise the human microbiome and contribute to both health and disease. The ability of Genesis to create computational models of cellular signaling and metabolism on its own should advance medicine, biotechnology, and fundamental biological knowledge, since such models are required to optimize experiments and interpret data to reveal the rules that govern biological processes. The Genesis project will involve three established research and training programs at Vanderbilt, all of which are active in the recruitment and involvement of undergraduate and graduate students and postdoctoral trainees in their research projects. Genesis offers a breadth of very attractive technical challenges and scientific and social opportunities ideal for interdisciplinary research training in AI, swarm robotics, machine learning, the exploration of signaling and metabolic pathways in microbes and suspended mammalian cells, and addressing how the complexity of biology could in fact be utilized to solve societal problems in nutrition, health, and medicine. Genesis will allow scientists and engineers and their trainees to address a number of important scientific, commercial, and societal problems by advancing our understanding of biology and disease and improving the efficiency of industrial production of biochemicals and pharmaceuticals. Genesis also provides additional impact as a tool for classroom instruction, in that ultimately it will allow students to pose questions and ask Genesis to design and conduct the experiments needed to answer them. Genesis will be designed to be mass produced at low cost, so that small laboratories could afford a small-scale system.In common use, yeast does its work in batches, where it grows and multiplies until it runs out of food or creates an environment where it can no longer thrive. A small batch of yeast grown in a research laboratory might require a milliliter of growth media in one well of a multi-well plate, whereas a yeast bioreactor at a pharmaceutical company could hold a few thousand liters, and one in a brewery a million liters. As an alternative to batches, a continuous-flow bioreactor, termed a chemostat, provides a steady supply of food and continuously removes excess yeast or even suspended mammalian cells and their metabolites to maintain steady-state growth. There is a growing recognition that chemostats can provide reproducible, reliable, and biologically homogeneous datasets that are well suited for probing the metabolism and signaling of living cells. However, the application of “self-driving” and machine-learning technologies to advancing biological knowledge will benefit from a thousand or more chemostats operating in parallel under computer control. Neither commercial production nor research chemostats have the correct combination of size, cost, and automated instrumentation. The Genesis system will address this need by using state-of-the-art, multi-channel microfluidic pumps and valves to control all of the microchemostats over a wide range of conditions with different strains of yeast, swarm robots to move 48 microchemostats at a time, and very high-throughput mass spectrometers to make a broad metabolic measurement every 10 or 15 seconds that will generate terabytes of data that exceed the ability of humans to control, process, and interpret. The resulting computational models could have thousands of equations. Genesis will provide, for the first time, an efficient means to design and conduct the massive number of biological experiments needed to parameterize, validate, and utilize these models to probe and even control biological systems for specific applications. Projects that will be pursued as soon as Genesis is operational include basic research in cell signaling and metabolism, quantitative explorations of the metabolomic interactions of co-cultured bacterial species that together could produce protein for food and chemical feedstocks, tracking multiple, parallel evolutionary histories to determine which environmental and genetic factors are important for the evolution of microbial cooperation, and improved methods to use mammalian cells to produce therapeutic antibodies. The parallel development of two Genesis instruments, one funded by NSF and the other by the Chalmers University of Technology in Gothenburg, Sweden, integrates Vanderbilt’s expertise in microfluidics and mass spectrometry with Chalmers’ expertise in AI, machine learning, and yeast to create a pair of robot scientists that will accelerate inquiries into the rules of life and the discovery of new solutions to some of society’s pressing problems.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.