Closed-loop learning of genome-scale models of metabolism using the "Robot Scientist" Genesis

One of the most challenging tasks facing 21st century science is the development of high-fidelity computational models of eukaryotic cells, for even simple eukaryotic cells, such as yeast, have thousands of different genes, proteins, and small-molecules, all interacting in complex temporal-spatial ways.

Developing these models is central to the future of medicine and biotechnology.

The necessary high complexity of the models means that developing and evaluating them will require the execution of many millions of hypothesis-led experiments.

Only AI systems, coupled with laboratory automation, have the ability to plan, execute, and record such a vast number of experiments.

In Chalmers I am developing ‘Genesis’, a 3rd-generation Robot Scientist, designed to automate the development of system biology models of yeast.

A Robot Scientist is a physically implemented laboratory automation system that exploits techniques from the field of AI to automatically execute cycles of scientific experimentation.

Genesis is designed to transform the automation of science: it will be able to execute in parallel 10000 continuous cycles of experiment. I will apply Genesis to improving the genome-scale modelling of yeast metabolism.

In the first part of the project Genesis will focus on piecemeal addition of kinetic parameters and changes to structure. In the second Genesis will investigate the addition of isozymes. During the project Genesis will execute 1 million hypothesis-led experiments.