The BRS team specializes in synthetic biology. Within this field, the team is designing and engineering metabolic pathways and genetic circuits for bioproduction, biosensing and biocomputation in whole-cell (E. coli) and cell-free systems.
The team is organized in a dry lab composed of data scientists and computational biologists, and a wet lab composed of biochemists and molecular biologists. Aside from its primary research goals (see research axes below), the team has been running a master program in Systems and Synthetic biology since 2010 at the University of Paris-Saclay, has coached iGEM teams (international competition in synthetic biology) since 2012 and has launched a spin-off private company in 2014 (Abolis Biotechnologies). The team is also engaged in community building in synthetic biology both in France and in Europe (CNRS GDR BIOSYNSYS, IRN SYNSYSBO, ELIXIR Microbial Community).
Funding sources: H2020 IBISBA, H2020 BioRoboost, BBSRC SYNBIOCHEM, PEPR B-BEST, ANR MEM, BASF
The most prolific application of synthetic biology and metabolic engineering is the bioproduction of molecules of industrial interest in host (chassis) microorganisms. To overcome the traditional lengthy and costly trial-and-error bioproduction process, the synthetic biology community has developed over the past decade the Design-Build-Test-Learn (DBTL) methodology.
When applied to bioproduction, the first DBTL step is the design of metabolic pathways enabling the synthesis of targeted molecules. That step is typically performed using retrosynthesis methods. With the RetroPath suite of softwares [1-3], the BRS team is internationally recognized in developing rule-based [1] retrosynthesis using beam-search algorithm [2] and reinforcement learning methods like Monte-Carlo Tree Search [3].
Many DBTL software pipelines and protocols developed by the synthetic biology community are available but do not follow FAIR principles. To palliate this shortcoming, we have compiled within the Galaxy workflow manager system tools for synthetic biology [4] (cf. Galaxy-SynBioCAD-video). The tools are used to build libraries of strains producing or sensing molecules of interest and engineer biosensors. The tools cover an end-to-end process, from (i) selecting using retrosynthesis the strains and targets to be synthesized/detected, (ii) designing the DNA parts to be assembled, to (iii) generating scripts driving robotic stations for plasmid assembly and strain transformation.
Funding sources: H2020 BioCellPhe, PIA3, iCFree, Equipex ALADIN, ANR SynBioDiag, ANR SINAPUV, DIM BioConvS
Cell-free is a fast-growing technology that is being used for bioproduction (like therapeutic and nutritional proteins) and medical and environmental biomarker detection. Despite its adoption by academics and industrials, cell-free expression systems still suffer from cost issues and batch to batch variation. To address these shortcomings, we have developed an AI based active learning methodology to reduce the cost of cell-free protein production and enhance reproducibility [5]. The method, fully automated, is integrated into our Galaxy platform, has been applied to optimize a CO2 carbon fixation cycle [6] and is currently used to produce antimicrobial peptides and proteins.
Whole-cell and cell-free biosensing are usually performed using transcription factors or riboswitches. Yet, the number of ligands directly activating or repressing transcription factors or riboswitches is limited. Rather than performing complex protein or RNA engineering we have developed a straightforward technology named SEMP (Sensing Enabling Metabolic Pathways) that makes use of enzymes to transform non-detectable ligands into detectable ones. Thanks to the large reservoir of enzymes, the SEMP technology enables one to increase the number of biosensors by several orders of magnitudes. Moreover, cell-free systems enable the detection of ligands that do not cross cell membranes (like proteins and RNAs), and cell-free biosensing devices are non-GMOs. The SEMP technology was therefore adapted to be cell-free, allowing multiplex detection of biomarkers used in phenylketonuria treatment [7] and prostate cancer diagnostic [8]).
Funding sources: HORIZON BIOS, ANR AMN, ANR DREAMY, ANR COSTXPRESS
Engineering computational devices is a long-standing endeavor of synthetic biology. While many genetic circuits have been built in the past to perform digital computations, little has been done to implement analog circuits. Mixing analog and digital circuitry, we have engineered the first perceptron (the basic unit of all neural networks) allowing us to classify samples based on their metabolic composition [9]. We have pursued this work establishing that in vivo native metabolic networks have the capacity to process information and to handle complex regression and classification tasks. To that end, we develop a new kind of hybrid neural network named Artificial Metabolic Network comprising a neural layer and a mechanistic layer [10].
Within the biocomputation research axis, the BRS team hosts an independent research group since 2024: the Cellular Computing Group (CCG). The group focuses on the engineering of genetic circuits, with increasing levels of complexity: going from the molecular level to multicellular circuits. By integrating experimental data with mathematical modeling, CCG aims to understand and predict the behavior of these circuits, facilitate their optimization, and scale them up.
Sources of funding below are National and European grants during the period 2018-2024. Other sources of funding include 4 PhD scholarships and 2 post-doctoral grants from INRAe, U. Paris Saclay, TWB, Genopole, and DGA.
