Advancing Health Data Exploitation through Secure Collaborative Learning
Recently, deep learning has been successfully applied in numerous domains and is increasingly being integrated into healthcare and clinical research. The ability to combine diverse data sources such as genomics and imaging enhances medical decision-making. Access to large and heterogeneous datasets is essential for improving model quality and predictive accuracy. Federated learning is currently developed to support this requirement offering an alternative by enabling decentralized model training while ensuring that raw data remains stored locally at the client side. Several open-source frameworks integrate secure computation protocols for federated learning but remains limited in its applicability to healthcare and raises issues related to data sovereignty. In this context, a French framework is currently developed by the CEA-LIST, introduces an edge-to-cloud federated learning architecture that incorporates end-to-end encryption, including fully homomorphic encryption (FHE) and resilience against adversarial threats. Through this framework, this project aims to deliver modular and secure federated learning components that foster further innovation in healthcare AI.
This project will focus on three core themes:
1) Deployment, monitoring and optimization of deep learning models within federated and decentralized learning solutions.
2) Integrating large models in collaborative learning.
3) Developing aggregation methods for non-IID situation.
Infertility is a growing problem in all developed countries. The standard methods for the diagnostic of male infertility examine the concentration, motility and morphological anomalies of individual sperm cells. However, 40% of male infertility cases remain unexplained with the standard diagnostic tools.
In this thesis, we will explore the possibility to determine the male infertility causes from the detailed analysis of 3D trajectories and morphology of sperms swimming freely in the environment mimicking the conditions in the female reproductive tract. For this challenging task, we will develop a dedicated microscope based on holography for fast imaging and tracking of individual sperm cells. Along with classical numerical methods, we will use up-to date artificial intelligence algorithms for improving the imaging quality as well as for analysis of multi-dimensional data.
Throughout the project we will closely collaborate with medical research institute (CHU/IAB) specialized in Assisted Reproductive Technologies (ART). We will be examining real patient samples in order to develop a new tool for male infertility diagnosis.
Development and multiparametric monitoring of a microfluidic chip of the blood-brain barrier model
The blood-brain barrier (BBB) protects the brain by controlling exchanges between the blood and nervous tissue. However, current models struggle to accurately reproduce its complexity. This thesis aims at developing and evaluating a microfluidic chip of BBB model incorporating a real-time monitoring system that combines simultaneous optical and electrical measurements. The device will enable the study of permeability, transendothelial resistance and cellular response to various pharmacological or toxic stimuli. By combining microtechnologies, cell co-cultures and integrated sensors, this model of biological avatar will offer a more physiological and dynamic approach than conventional in vitro systems to improve understanding of the diffusion/permeation phenomena of therapeutic molecules. This project will contribute to the development of predictive tools for neuropharmacology, toxicology and research into neurodegenerative diseases.
Structural snapshots of a substrate within the active site of a mitogen-activated protein kinase
Mitogen-activated protein kinases (MAPKs) are key signaling enzymes that regulate cellular stress responses through the phosphorylation of specific protein substrates. Dysregulation of MAPK signaling contributes to numerous diseases, including cancer and neurodegenerative disorders. Although MAPK activation and catalytic mechanisms are well characterized, the structural basis of substrate specificity remains unknown. This project aims to address this gap by capturing atomic-level structural snapshots of substrates bound within the active site of the c-Jun N-terminal kinase 1 (JNK1). To achieve this, we will employ X-ray crystallography together with innovative nuclear magnetic resonance (NMR) methods that integrate selective methyl isotope labeling and photoactivatable catalysis. By elucidating the structural details of how substrates are recognized by the active site of JNK1, our work will open new avenues for the development of substrate-competitive inhibitors of MAPKs with enhanced selectivity and therapeutic potential.
Adaptation and degradation of PFAS by the bacterium Pseudomonas putida
Per- and polyfluoroalkyl substances (PFASs) are a class of very diverse chemicals found in products of daily use, that are highly persistent and encountered everywhere in the environment. They accumulate/biomagnify within the natural food chain and show a relatively high toxicity including the alternative products developed after the ban of the legacy compounds. Therefore, the world is facing a situation of great concern all the more as the retreatment of contaminated soils, sediments and water is difficult and costly. One of the main challenges is because various PFASs have quite different physicochemical properties but are often encountered in mixture making it difficult to find a technology efficient to remove all of them. We propose to pave the way towards another approach for PFASs elimination, bioremediation that is known to be a good alternative to chemical or physical methods for removing toxics (self-sustainability, cheaper, working in milder conditions, and often with dissolved and sorbed contaminants). A few bacteria have been described to be able to partially modify/degrade some PFASs. However, except the aspect of PFAS transformation, no data are available concerning their adaptation to PFAS exposure. A few projects are focusing on finding enzymes implicated in the degradation per se but if we want to use bacterial cultures and not enzymes, many other parameters need to be taken into account to set up a performant strain and hence a performant process. Therefore, we propose to analyze in depth the response to several PFASs of the PFAS degrading strain Pseudomonas putida ATCC 17514 in term of degradation, adaptation to a potential toxicity and metabolism adjustment. The analyses will mainly rely on a proteomic approach that is a very powerful technique to analyze global responses without a priori, and has never been done to characterize PFASs toxicity or fluorinated compounds metabolism in bacteria. The ultimate goal after this bootstrap project will be to engineer or select a robust and efficient strain capable of biodegrading PFASs.
Development of monoclonal antibodies for the diagnostic and the treatement of hypervirulent-Klebsiella pneumoniae
For several years, we have observed the emergence of hypervirulent (hvKp) strains of Klebsiella pneumoniae that have become highly resistant to antibiotics. In a context of dwindling antibiotic options, monoclonal antibodies (Abs) directed against well-conserved capsular antigens of these hvKp strains appear as a promising therapeutic alternative.
This PhD project is structured around three complementary objectives:
1. To describe the circulation of hvKp clones through comparative genomic analysis of strains collected via the French National Reference Center for Antibiotic Resistance and through an international collaboration.
2. To produce and characterize monoclonal Abs directed against the HvKp capsule.
3. To develop a rapid detection tool based on MALDI-TOF profile analysis coupled with machine learning algorithms.
Acellular Biotherapy with Optimized Immunomodulatory Properties for the Prevention of Organ Injury in Traumatic Contexts
Severe trauma causes more than 5.8 million deaths worldwide each year, often associated with massive hemorrhages and multiple organ failure (approximately 33% of cases). Rhabdomyolysis, common in these patients, results from the destruction of muscle cells and leads to the release of their contents into the bloodstream. This complication promotes acute kidney injury and liver dysfunction. Currently, no specific treatment exists; management remains primarily symptomatic. Mesenchymal stromal cells (MSCs) are widely used for their immunomodulatory and regenerative properties. Preclinical studies have shown that IL-1ß-preconditioned MSCs can prevent kidney and liver damage and reduce vascular permeability after hemorrhagic shock. Their efficacy relies on the secretion of soluble factors and extracellular vesicles, known as acellular products. A large-scale, clinical-grade production method for these products, based on tangential flow filtration, has been developed. These products exhibit experimentally demonstrated immunomodulatory activity and hepatoprotective effects. Ready to use and easy to store, they represent a promising alternative to cell therapies in emergency settings. The objective of this thesis is to optimize the immunomodulatory and anti-inflammatory properties of these cell-free products by promoting their expression of two key immune tolerance molecules, PD-L1 and HLA-G. We will evaluate the interactions between these optimized products and various immune cells in vitro, and then in vivo in a traumatic hemorrhagic shock model (rat model).
Influence of Cytomegalovirus on Tissue-Specific Immune Responses in Non-Human Primate
Most studies in anti-infectious immunity focus on characterizing pathogen-specific immune responses and identifying strategies to optimize them. It is now essential to consider interindividual variability related to age, sex, metabolic status, and infectious history, which strongly influence these responses.
IDMIT’s expertise in preclinical modeling of viral infections provides an ideal framework to address these questions. Cytomegalovirus (CMV) infection represents a relevant model due to its high prevalence, its age-dependent effects, and its association with immune aging. Although epidemiological data suggest that CMV seroprevalence impacts responses to other infections and to vaccination, the underlying mechanisms remain poorly understood. We hypothesize heterogeneous effects related to the diversity of host–virus interactions across sites of viral persistence.
This project aims to characterize CMV-specific immune responses in blood and tissues of young and aged non-human primates, and in the context of chronic SIV infection. The objectives are (i) to assess age-related differences in viral dissemination and immune responses, (ii) to evaluate the predictive value of blood markers relative to tissue parameters, and (iii) to study the reciprocal modulation of CMV and SIV responses during co-infection.
These studies will contribute to the development of vaccination strategies targeting the deleterious effects of CMV and the tissue-specific modulation of immune responses.
Active matter: self-organization of mitotic spindles
The mitotic spindle is an essential cytoskeletal structure that enables chromosome separation during cell division. This project seeks to identify the physical principles that control spindle assembly by using a simplified biomimetic system composed solely of microtubules and molecular motors. We will use motors of opposite polarities combined with dynamic microtubules to understand how these components organize through active phase separation. Indeed, preliminary experiments have demonstrated that such reconstituted systems can spontaneously form bipolar structures resembling mitotic spindles. We now propose to encapsulate these molecular components in compartments of controlled geometry to reconstruct a minimal bipolar structure capable of elongating, retracting, and separating its organizing poles. This multidisciplinary approach will combine biochemical and physicochemical techniques, advanced microscopy, and quantitative analysis of the spatial and temporal evolution of the system. The experimental work will be closely coupled with theoretical modeling in collaboration with Prof. Jean-François Joanny (Collège de France) to develop a physical model of active phase separation that will provide better understanding of self-organization mechanisms at the subcellular scale in living organisms.
Origins and evolution of prion-like proteins (PrLPs) in eukaryotes
Initially associated with neurodegenerative diseases, prion-like proteins (PrLPs) are now recognized as key physiological players in cellular plasticity and stress response. These proteins often contain an intrinsically disordered domain rich in glutamine and asparagine, known as a prion-like domain (PrLD), capable of switching between soluble, condensed, or amyloid states. Notable examples include CPEB in Aplysia, involved in synaptic memory, MAVS in antiviral defense, MED15 and FUS in transcriptional regulation and nucleocytoplasmic condensate dynamics, and ELF3 in plants, whose amyloid polymerization controls flowering and photoperiodic responses. In fungi, Sup35, Ure2p, and HET-s serve as experimental models of functional prions, demonstrating that reversible aggregation can act as a regulatory or adaptive mechanism. These conformational transitions are now viewed as adaptive molecular strategies rather than pathological anomalies.
This PhD project aims to trace the origin and diversification of prion-like proteins across eukaryotes, testing the hypothesis that major paleoclimatic crises have episodically promoted the emergence and duplication of genes encoding PrLDs through microsatellite expansion and transposable element activity. The project will combine large-scale phylogenomic analyses, PrLD domain detection, and modeling of selective pressures to map the key stages in the functional evolution of PrLPs and their links to stress tolerance.