Our lab aims to develop new general methodologies enabling future scientists to engineer living cells with completely new capabilities de novo. We want to restrict ourselves to two main applications: 1) de novo engineering of organism-specific antibiotics, and 2) de novo engineering of living artificial intelligence. 

For the first application, we use directed evolution of phage-like particles. Although, because we do it de novo, there is no initial activity we could use for selection. As this is mandatory in directed evolution techniques, we circumvent this requirement. We aim at the engineering of antimicrobials devoid of any DNA without requiring to even culture the pathogen, we only require its genomic data. Could we one day have a machine that generates personalised on-demand antimicrobials once we pour in it a sample of our infection? Current approaches rely on trial-and-error techniques where new molecules (chemically synthesised from libraries or isolated from nature) are sampled against the targeted pathogen. This takes many years, billions of Euros and bacteria becomes resistant to them in a few hours by using evolution. We aim at using the same weapons than bacteria in this antibiotic-resistance war.

For the second application, we have created a new type of genetic circuits with analogue memory able to be reprogramed by reinforcement learning. In fact, the reinforcement learning acts as a directed evolution except for the fact that it doesn’t use mutations or genetic inheritance. We aim at engineering microorganisms and synthetic tissues able of complex computation by engineering in the artificial neural networks. For instance, we have recently engineer E. coli cultures to be able to learn playing board games (simplified variants of tic-tac-toe, chess and Go). Could we one day have a computer that is alive? Most artificial intelligent approaches rely on doing deep learning on electronic computers, but our brain can still do it better using only 20 watts of power consumption. How can we outperform it?

We developed automated design techniques able to engineer RNA-based sensors to discriminate specific cell and cell-states.