Synthetic biology has a large potential to tackle the issues of circular economy, from converting waste to useful products to carbon-neutral industrial production, through the use of synthetically engineered microbes. These engineering efforts can potentially be expedited with AI-assisted design processes, however, this potential still largely unharnessed. We envision a AI system facilitating the DBTL cycle (Design, Build, Test, Learn), where the strain is designed (D), built in a laboratory (B), measured and tested (T), to learn (L) a model on the current strain to be exploited for the next cycle design phase (D) again. Automatic operation of the DBTL loop has so far been been demonstrated only in selected settings.
In this project, we propose to develop new AI approaches to tackle the problem of accelerating the design of synthetic microbial strains. In short, we propose learning a model to suggest modifications to an existing design by a reinforcement learning approach, where the feedback from the testing of a design is used to propose new, improved designs. The starting point is a reinforcement learning approach called Actor-Critic model, where the Actor component models and updates the policy (what actions to choose in given state) and the Critic component models the value function (goodness of the current state of the system). Such a model has been recently shown to be promising approach for finding genetic modifications that maximise the productivity of synthetic microbial strain. The project will be conducted in collaboration with Synthetic biology team at VTT Technical Research Center of Finland.
The ideal background for the student is an MSc in computer science, mathematics or statistics, with strong experience in machine learning. Experience in reinforcement learning, robotics, control systems engineering, synthetic biology or systems biology is considered as an advantage.
Fill an application form at www.hict.fi/spring2020 by the call deadline Mon, 27.01.2020