Synthetic Biology · Software · Chemical Reaction Networks · Grand Challenges

William Poole

Questions? Ask them on our slack or anonymously through our form!

Having trouble listening to the podcast above, or want to take it on the go? Try downloading it here.

Show Notes

This week we spoke with William Poole, a graduate student at Caltech working on quite a few topics! His research spans synthetic/systems biology to molecular programming, software development to chemical reaction network (CRN) theory, machine learning to cell free systems. We certainly had a lot to talk about!

We started off by discussing BioCRNPyler, a library which Will has been working on that allows for the rapid development and compilation of complex CRNs. He describes how BioCRNPyler can help you rapidly design CRNs in a variety of cellular contexts. The CRNs can then be simulated using any simulator/solver. We also discuss other software projects he is involved with such as Bioscrape and Vivarium.

Next we move onto William’s research into chemical Boltzmann machines, what they are and how they are related to machine learning, while talking about how low molecular copy number systems might be able to perform more complex computation than high copy number systems.

We also talk about how William got into molecular programming from his undergraduate degree, which focussed on physics and biology. He describes how his undergraduate research led him in various directions, and even into working in bioinformatics at the Institute of Systems Biology for a few years before pursuing graduate school.

This ultimately spurred on a somewhat grand discussion on William’s “dream” for molecular programming. He is very concerned about climate change, and talks at length about how in the long term we might be able to program many of the materials around us to sequester carbon, and eventually “re-terraform” the earth. Finally, we asked why physicists and engineers are able to come together to build large scale projects such as the LHC and ISS, while no such projects exist for the biological sciences, and we speculate on what such a project could look like for our field...


I received my B.Sc. from Brown University in Biological Physics and am scheduled to complete my PhD in Computation and Neural Systems from Caltech in summer 2021 (co-advised by Erik Winfree and Richard Murray). Broadly, my research interests involve developing mathematical and computational tools to understand “how cells think” and “how to program cells”. I view systems biology and synthetic biology as two sides of the same coin; to truly understand and control biological systems we will need fundamentally new ways of thinking about biochemical computation. For inspiration, I have looked to statistical physics, machine learning, and computer science. I also believe that doing theory in a vacuum, without close contact to experimentalists, is counterproductive towards developing the tools that will drive science and technology forward. Towards this end, I also dabble in the wet lab side of synthetic biology with an eye towards applications in green technology and sustainability.


I have worked to enable the design and analysis of biochemical networks through techniques inspired by machine learning. Recent work implements specific machine learning algorithms with biochemical models and uses ideas from statistical physics and information theory to interpret broad classes of biochemical networks as machine learning algorithms.


I have been inspired by the machine learning libraries like pytorch and tensorflow which provide abstract frameworks for specifying and efficiently training diverse machine learning architectures. In order to successfully engineer and understand the complexity inherent in any biological system, we will need powerful computer aided design tools. This means the ability to easily compile diverse models of biochemical circuits (BioCRNpyler GitHub), fast simulation techniques custom-tailored to utilize and learn from biological data (bioscrape GitHub), and the ability to connect, combine, and simulate diverse models across many scales (Vivarium Github).


I have worked on understanding and modeling the metabolism of E. coli cell extract. It amazes me that in this “not even living” biological system we still have only barely begun to understand how to model, engineer, and optimize metabolic activity. Ultimately, I see metabolic engineering as a cornerstone of a green-economy which produces all kinds of chemicals and materials from living matter instead of fossil fuels.