Wikswo and VIIBRE team on track to build third-generation ‘self-driving lab’ with $1M from National Science Foundation

John Wikswo, founder and director of the Vanderbilt Institute for Integrative Biosystems Research and Education and Gordon A. Cain University Professor, is the principal investigator of a $1 million award from the National Science Foundation. The object is to build a pathbreaking “robot scientist”—a fully automated microfluidic system for parallel, independent, long-duration, machine-guided experiments. The target organisms are the single-cell eukaryotic yeast Saccharomyces cerevisiae, commonly known as brewer’s and baker’s yeast, the bacterium Escherichia coli and other microbes used in commercial biotechnologies, and the Chinese hamster ovary (CHO) cells used to produce antibody-based drugs and vaccines. 

John Wikswo
John Wikswo

Wikswo had previously developed an award-winning MultiWell MicroFormulator to recreate in a plastic, 96-well plate the time-dependent drug concentrations that previously could be replicated only in animals. In 2020, Ross King, a professor of machine intelligence at Chalmers University of Technology in Gothenberg, Sweden, and the developer of Adam and Eve, the first robot scientists, recruited Wikswo to convert the MicroFormulator into a third-generation robot scientist, Genesis, to provide King’s group with thousands of miniature, self-driving chemostats that will be used to understand yeast metabolism and signaling. Realizing that a self-driving biological laboratory has many other applications, Wikswo sought NSF funds to create a Genesis system at Vanderbilt that could be used not only for fundamental studies of yeast but also of the breadth of cells used in biotechnologies.

The vast complexity of biology is legendary: humans have about 20,000 genes, between 80,000 and 400,000 proteins, and millions of metabolites produced by different chemical reactions, such that the creation of a mathematical, systems biology model of all of the parts presents a daunting challenge. To simplify the problem, the focus of the robot’s initial experiments will be on S. cerevisiae, which has only 6,275 genes and is widely studied by biologists due to its similarities to many mammalian systems. Very large libraries of strains of S. cerevisiae have been created with one or more of these genes inactivated or over-expressed, and most biological processes involve multiple genes. Clearly, the number of possible experiments with even this simple, single-cell organism far exceeds the capabilities of human scientists using standard laboratory techniques. Hence the need for a robot scientist. 

Genesis will merge cutting-edge microfluidic and mass spectrometry technologies developed at Vanderbilt with AI software developed by the King group at Chalmers that formulates a biological hypothesis, then selects, designs, conducts and analyzes experiments, refines the mathematical model for the microbe and repeats the entire process. At each step, Genesis will select which yeast strain to use, what to feed that strain and the temperature, pH, oxygen level and stirring speed for that specific experiment. The appeal of Genesis is that it will be able to conduct a thousand or more independent experiments simultaneously. 

“For the first time, we will be able to study hundreds of different drug profiles at the same time,” Wikswo said. “While simultaneously running all of these experiments, the artificial intelligence in Genesis will be formulating a scientific hypothesis based on a computer representation of a biological model.” 

At the heart of Genesis are VIIBRE-developed microfluidic pumps and valves and the microcontrollers and software that operate them. In contrast to making bread, brewing beer or growing yeast in the earlier Adam and Eve robots, where the yeast are fed at the beginning of the process and allowed to consume that food, possibly until it is gone, a chemostat provides a continuous flow of nutrients and washes away waste products and whatever yeast don’t fit within the chemostat. The steady state that occurs in a chemostat is much easier to understand than the continuously changing levels of nutrients, waste products and gene expression that occurs throughout any batch process, which is why King approached Wikswo. 

For most commercial chemostats, only a small number of measurements are made, for example pH, glucose concentrations and oxygenation. In contrast, co-principal investigator John McLean, Stevenson Professor of Chemistry and expert in state-of-the-art mass spectrometry, will teach Genesis how to quantify changes in large numbers of metabolites, thereby giving King vast amounts of data that he can feed to his robot-scientist algorithms. Genesis will then be able to perform the myriad measurements it needs to illuminate currently unknown microbial metabolic and signaling pathways. 

Contributing to the yeast biology research and chemostat design and operation are Eric Spivey, co-principal investigator and research assistant professor of biomedical engineering at Vanderbilt, and Ievgeniia Tiukova, a researcher in the Department of Biology and Biological Engineering at Chalmers. Ron Reiserer, VIIBRE’s senior research and development engineer, is the chief designer of the Genesis platform, and physics graduate student Kyle Hawkins is modeling both Genesis chemostats and yeast biology. Mark Styczynski and Eberhard Voit, Genesis users at Georgia Tech, will investigate automation of metabolomic data extraction and model design for S. cerevisiae and repurposed gut communities.  

Ron Reiserer
Ron Reiserer works with a prototype in the lab at Stevenson Center. (John Russell/Vanderbilt University)

Participating in the exploration of other microbial systems are co-principal investigator Megan Behringer, assistant professor of biological sciences, who will use Genesis to explore interacting bacterial populations, as will Michael Lynch at Arizona State University. Jamey Young, Cornelius Vanderbilt Professor of Engineering, will explore how Genesis might improve CHO cell production of antibodies. Michael Henson of the University of Massachusetts at Amherst plans to use Genesis to study how mixed populations of human gut microbes can be used to convert sugars to industrial chemicals.  

Wikswo has already shown that the Genesis project provides ample opportunities for undergraduate and graduate students, for example as a central topic of his Systems Biology seminar during the Fall 2021 semester. 

The iterative hardware development project funded by both Chalmers and NSF will require continuous collaboration to refine Genesis as it evolves. “We’re working together to create an instrument in a class that really has not been created before,” Wikswo said. “What is missing from computational models of biotechnology and medicine is an effective means to design and conduct the massive numbers of experiments needed to utilize these models for specific applications. That’s Genesis.” 

Wikswo envisions the many future applications of Genesis beyond experiments on single cell organisms. “Ultimately, Genesis will be able to take incredible data from everything from yeast and bacteria to organ chips to cultured cells to maybe even zebrafish,” Wikswo said. “It will become a general-purpose perfusion, control and analysis platform to ask questions and explore science, so it may be applied to some of the world’s most pressing problems.”  

The Genesis project is being built around a Vanderbilt-grown technological base—Wikswo’s 40th patent was issued in early October 2021; more than half of these are the product of VIIBRE research now in its 20th year.