Conception assistée par IA de nanoparticules lipidiques encapsulant des polynucléotides

How can the right manufacturing pathway be identified to obtain the desired functional structure in a multiparametric space? A combination of physical chemistry and process engineering, supported by microfluidics and artificial intelligence.microfluidics and artificial intelligence.

Scientific context

Lipid nanoparticles (LNPs) are now the leading non-viral vectors for RNA and other polynucleotides. Yet their development remains largely empirical: two formulations with similar sizes may differ in their internal organization, loading state, or stability. This PhD project will seek to connect, within a single framework, composition, process, structure, and function. A major challenge is the size of the parameter space, which includes numerous composition and process variables and is compounded by the cost of the polynucleotides used. Addressing it requires precise control over how the different molecular species are brought into contact, using microfluidics; characterization of the resulting structures by a range of scattering and microscopy techniques; and robust, resource-efficient strategies for exploring the parameter space through artificial intelligence.

The PhD project

• Develop a multi-chip mode: by combining several rapid-mixing chips, the parameter space can be explored with additional degrees of freedom.

• Combine fast and structural measurements: online DLS/SLS and transmission, pH and encapsulation measurements on collected aliquots, followed by SAXS, SANS, or cryo-TEM on the most informative samples.

• Explore a complex experimental space by varying composition and process parameters, including pH, water/ethanol ratio, mixing speed, N/P ratio, PEG-lipid content, and the nature or length of the polynucleotide.

• Use active learning and multi-fidelity models to select subsequent experiments, reduce material consumption, and generate explainable models.

Originality and expected outcomes

The originality of the project lies in treating the process as a design variable rather than as a simple manufacturing step.

The PhD researcher will develop a system combining several rapid-mixing modules with optical detectors. It will build on the existing platform, extend it to multiple successive operations, and couple it with other characterization techniques. A database will be assembled, and interpretable predictive models will be developed by combining parametric exploration with AI-assisted exploration. Rules linking composition, process, and structure will thus be established.

Research environment

The project will be carried out at the Laboratoire de Génie Chimique in Toulouse, within the Colloids and Complex Fluids team. It will use the 2FAST platform of the DIADEM Discovery Hub. An instrumentation partnership with Cordouan Technologies is planned, together with a collaboration with the Nordic COMMONS consortium (Lund University, University of Copenhagen, KTH Royal Institute of Technology, and Chalmers University of Technology in Gothenburg).

Candidate profile

A Master’s degree (M2) or engineering degree in physical chemistry, chemical engineering, materials science, nanoscience, pharmacy/biophysics, or a related field is required. Curiosity, an interest in experimental work, rigor, and enthusiasm for data analysis are essential. Experience in microfluidics, colloids, scattering techniques, or Python/ML is welcome, but motivation, willingness to learn, and critical thinking will be the main selection criteria.

CONTACT AND APPLICATION

Dr Kevin Roger, CNRS researcher — kevin.roger@cnrs.fr

Initial application: CV and a few lines outlining your background and motivation.

More info here