"What you can not create, you can not understand.", Richard Feynman
The book 'What is life' by Professor Erwin Schrodinger invigorated my scientific interest in synthetic biology. I was motivated to learn about how Professor Schrodinger applied physical reasoning to biological systems, in the year 1944 when the fields 'synthetic biology' and 'systems biology' were non-existent. The explanation on how ordered biological systems arise from disordered systems interested me and I started pondering on how I can decipher biological systems using my engineering skills.
The last several years have witnessed the emergence of synthetic biology, a new area of biological research that aims to engineer biological systems with desired functions for broad applications. This line of research has led to the development of diverse systems, such as oscillator, communication-based circuits, band detectors, bistable switches, ribo-switches, logic gates , and drug-producing cells.
Synthetic biology relies heavily on systematic design cycles. Given a conceptual design, mathematical modeling is often used to explore dynamics of the design, which may offer guidance for subsequent implementation, test, and revision. For gene circuits with complex dynamics, modeling becomes more critical. In addition to suggesting choices of appropriate circuit components to initiate implementation, modeling may also identify critical components for further optimization by rational design or directed evolution. Conversely, data collected from experimental tests can provide valuable information for evaluating and revising mathematical models and kinetic parameters.
However, the design cycles have focused primarily on steady-state dynamics of gene circuits in constant host environments, which provide resources for gene expression such as ribosomes, polymerases, and energy. The approach makes intuitive sense from engineers' perspective: to isolate host physiology from circuit design and to reduce system complexity. This approach, however, faces two challenges: fluctuating host environments can profoundly impact functions of synthetic gene circuits18; in fluctuating host environments, dynamics of synthetic gene circuits need to be tightly controlled to achieve optimal performance. As such, fluctuating host environments can significantly reduce predictability of synthetic biological systems. This often results in laborious cycles of trial-and-error during the optimization of synthetic biological systems. Can we predict the impact of fluctuating host environments on synthetic gene circuits?
My career goals are to improve the rational engineering of synthetic biological systems by three ways. First, I will construct artificial cells using a bottom-up approach, which emulates natural cells by enclosing protein expression systems in lyposomes. This way, I could control host environments of synthetic gene circuits, thereby improving predictability of synthetic cells. Second, I will analyze circuit-host interactions by using both mathematical modeling and experiments. Specifically, I aim to unravel design principles of gene circuits in fluctuating host environments. The design principles would be generic and applicable for the engineering of other synthetic biological systems. Third, after establishing both the mathematical and experimental foundations, I will construct active learning machines to assist design cycles of both artificial and natural cells. To fundamentally improve the design cycles of synthetic biological systems, the active learning machines need to fulfill at least four criteria. First, the machines should accumulate knowledge by mining literature data and by using experimental characterization of synthetic biological systems. Second, when being posed specific design objectives, the machines should be able to suggest experiments or propose new hypothesis. Third, the machines should help to manage complexity of synthetic biological systems, especially circuit-host interactions. Fourth, the machines should be able to handle and update multi-scale models. Indeed, preliminary forms of the machines have been applied to generate hypothesis for functional genomics and electrochemistry experiments.