Designing optimal feedback control of stochastic gene expression: an interdisciplinary approach using control theory, statistical physics and machine learning

Controlling gene regulatory networks can be a fundamental step for their use in the synthesis of therapeutic strategies as, for example, to correct the severe phase lag observed in the circadian rhythms of some patients. However the stochastic nature of gene expression [1] makes the effective implementation of feedback control strategies a non-trivial problem. Computer-guided control of gene expression to achieve a desired level of mean protein numbers has only recently been demonstrated in the lab [2,3]. In this project the student will combine ideas from statistical physics and control theory to design a new controller that can simultaneously control both the mean number of proteins and the size of the fluctuations about this mean. Specifically the student will use recently developed accurate approximation methods [4] to implement stochastic model predictive control of gene expression in eukaryotic cells and to test this strategy using stochastic simulations. In parallel, the student will also explore the use of artificial neural networks to achieve the same aim. The simulated controller will then be tested in the laboratory; successive rounds of modelling and experiment will be used to build a fine-tuned efficient controller.

5. Further Information
This project is ideal for a student with a Bachelors or Masters in Applied Mathematics, Physics, Computer Science, Engineering or a closely related field. Previous familiarity with mathematical modelling in biology is useful but not a necessity. The student will be given extensive training in stochastic modelling, control theory and machine learning in the context of biology in their first year to ensure a solid foundation. The student will be part of the group of Dr. Ramon Grima http://grimagroup.bio.ed.ac.uk/index.html, which is located in the Centre for Synthetic and Systems Biology (SynthSys) at the University of Edinburgh. The experiments will be conducted in the lab of Dr. Filippo Menolascina.