CV
Biotechnologies hold the potential to address global challenges in agriculture, healthcare, energy, and manufacturing. Traditionally, these fields have relied on large-scale design and experimental campaigns, but advances in control theory and machine learning, alongside lab automation and synthetic biology, open new opportunities for innovation. My goal is to directly interface learning and control algorithms with living organisms to establish new and more efficient avenues of biological research and engineering.
Although machine learning has driven recent breakthroughs in biology, these often depend on supervised learning, which requires large, diverse, and labeled datasets. I envision that interactive data-driven frameworks that “seek” new knowledge instead of passively fitting pre-acquired data will be critical to the future of engineering biology: They will allow for high-throughput and precise control of cellular processes, enable multi-scale and multi-modal interactions with biology, and will make it possible to efficiently traverse complex design and input landscapes.
These frameworks will require interactive loops between cells and computers. My research group will capitalize on my expertise at the intersection of synthetic biology, control theory, and machine learning to transfer data-driven approaches to biological research and engineering: I am interested in a broad range of questions and problems including optogenetic control of growth in single cells for biomanufacturing applications, end-to-end reinforcement learning for the spatial organization of microcolonies, and active learning and robotics for the optimization of expression vectors or synthetic circuits.
I have developed real-time control interfaces to drive gene expression in real time and in single cells, for example to control a genetic toggle switch around its unstable equilibrium point. More recently, I have leveraged modern data-driven algorithms to control thousands of single cells in parallel with optogenetics, and applied this method to drive and study the dynamics of antibiotics resistance. Since then I have been working on transferring methods from reinforcement learning to cell control problems, in particular to drive cell growth in real time with light.
Lugagne et al. 2024 Klumpe, Lugagne, et al. 2023 Lugagne, et al. 2017
In 2019 I created DeLTA, a software suite that uses deep learning to rapidly and robustly segment, track, and reconstruct lineages in microscopy time-lapse movies. Since then DeLTA has been adopted by many labs worldwide and others have joined me in my effort to keep improving it.
Single-cell gene expression dynamics are critical to understanding bet-hedging strategies in cell response. I have used my expertise in microfluidics, single-cell microscopy, and data analysis to quantify the impact of gene expression dynamics and cell-to-cell heterogeneity on antibiotic resistance in bacteria.
I have developed bioproduction strains in collaboration with the Cheng lab at Boston University to evaluate stimulated Raman spectro-microscopy methodologies for label-free imaging of metabolites of interest such as fatty acids or terpenes in single cells. This technique not only allows us to quantify concentrations and the spatial organization of bioproduction, but also the chemical composition of the products for differently optimized strains.
†: co-corresponding author
*: co-first author
See my Google Scholar page for the complete list.
I co-organize the Biocontrol seminars series, a series of monthly online seminars at the intersection of control theory and biology. Feel free to reach out if you would like to speak!
For questions regarding DeLTA check our GitLab repository, in particular the issues system. See also the online documentation.
jlugagne [at] bu [dot] edu
jean-baptiste [dot] lugagne [at] eng [dot] ox [dot] ac [dot] uk