Machine learning for visual inspection and final state reconstruction in the Atlas experiment
The Large Hadron Collider (LHC) is taking data at an unprecedented energy of 13.6 TeV, allowing to explore a new energy frontier and rare processes. In the meantime its high luminosity phase (HL-LHC) is actively being prepared. One of the main goals of both the on-going run and the HL-LHC phase is to scrutinize potential deviations from the predictions of the Standard Model of particle physics. For that purpose, it is particularly relevant to study the cornerstone of the model, the Higgs boson, as well as the heaviest known elementary particle, the top quark.
The PhD topic is aiming to develop machine learning techniques for two different challenging projects within the ATLAS experiment at the LHC: visual inspection of new ATLAS tracker modules, and reconstruction of final state objects to study rare processes with multiple leptons in the final state.
In order to meet the performance required by HL-LHC, the current ATLAS inner detector will be replaced by a new all-silicon detector, partially composed of pixel modules. The pixel module production includes multiple assembly steps each one being followed by quality control. The first part of the PhD project is aiming to automate visual inspection of the module metallic contacts (pads) before connecting them to the electronics. Initial work showed that visible defects presented on the pad surface are detectable by machine learning algorithms.
The PhD work will explore several directions:
- Full investigation of supervised approaches. The production start-up period will allow to collect the dataset necessary to train supervised segmentation models targeting not only anomaly detection, but also their classification.
- Investigation of unsupervised anomaly detection based on generative models. The DRAEM model [1] for instance learns a joint representation of an anomalous image and its anomaly-free reconstruction, while simultaneously learning to distinguish them. The reconstructive sub-network is formulated as an encoder-decoder architecture and can be replaced with more advanced types of generative models such as GANs.
- Investigation of various models treating an anomaly-free pad surface as a background noise. Signal processing denoising algorithms can then be used to increase the information content while e.g denoising autoencoders can learn features that are invariant to noise.
The second part of the PhD will focus on rare processes involving top quarks with bosons (ttW, ttZ, ttH, tH, 4tops). These processes have just been observed for the first time using the dataset recorded at 13 TeV (except for tH) in final states with several leptons. However these analyses have always been carried out separately, which strongly limits our overall physics understanding, the correlations between the various measurements being large and unknown. The new approach proposed here consists of designing a global analysis to tackle these limitations. This approach requires cutting edge techniques to reconstruct precisely the final state. Indeed several degrees of freedom originate from the undetected neutrinos and combinatorics is made complex by the multiple decay products. An approach based on GNN was already tried. Although it provides a certain level of accuracy, some other approaches have to be explored to improve the reconstruction performance. There are in particular two models of interest developed within the LHC community:
- The topograph model [2], also based on GNN, makes use of additional information about the event decay chain.
- The Spanet model [3], a symmetry-preserving attention network exploits natural invariances to efficiently find particle assignments without evaluating all permutations.
An assessment of the relevant uncertainties is also an important aspect when using these tools.
This dual project provides an opportunity to be involved in the full chain of the AI application development, being also part of the world-wide ATLAS collaboration.
Turbulent geophysical flows above topography
The project is focused on the large-scale organization and the energy budget of idealized oceanic flows above random topography. We will first characterize the enhanced energy dissipation due to the topography and its impact on the transport properties of an idealized parcelle of Ocean. To wit, we will consider models of increasing complexity, from quasi-two-dimensional to fully three-dimensional. We will then turn to the large-scale spatial organization of the flow, building on recent progress regarding the application of statistical mechanics to such flows. A special effort will be devoted to adapting these tools to forced-dissipative flows.
Towards real-time simulation of thermal scenes in a tokamak to support plasma operations.
Monitoring the surface temperatures and heat fluxes of the walls in nuclear fusion devices is crucial for the operation of fusion machines. To ensure the reliability of these measurements, particularly through infrared imaging, CEA is developing a digital twin capable of modeling the entire infrared (IR) measurement chain, from the thermal source to the sensor.
The objective of this thesis is to create a thermal model that can predict heat fluxes and surface temperatures across the entire machine wall, with a goal of real-time computation. This approach is based on two key developments:
1)Development of a Monte Carlo statistical method: This method will solve the heat equation over large geometries in a complex environment, including a variety of heat sources and materials.
2)Acceleration of calculations on graphics processing units (GPU): Utilization of the Kokkos environment to optimize calculation performance while ensuring portability across all high-performance computing (HPC) platforms.
These developments will be validated and quantitatively evaluated on two experimental platforms: the laboratory test bench MAGRYT and the WEST tokamak, used as a demonstrator machine. The thesis will be conducted in a collaborative framework between CEA/DRF/IRFM and CEA/DES/ISAS. The developments will be integrated into the IR digital twin developed by CEA/IRFM for fusion machines and within a dedicated ray-tracing application for CEA/DES.
Development of novel therapeutic approaches for the treatment of Friedreich's ataxia targeting the biochemical process of iron-sulfur cluster assembly
Friedreich's ataxia (FA) is the most common hereditary ataxia, with 1 in 30,000 people affected worldwide. It is a genetic, neurodegenerative and cardiac disease caused by defective expression of frataxin (FXN), a mitochondrial protein stimulating the biosynthesis of iron-sulfur (Fe-S) clusters, which are metallo-cofactors of proteins involved in a multitude of essential biological functions. The thesis project aims to develop drugs for the treatment of this disease by combining biochemical and biophysical techniques, in vitro screening and tests in animal models.
We have previously shown that FXN acts by stimulating the supply of sulfur to the Fe-S cluster biosynthesis system. More recently, we have discovered and extremely fine cross-regulation between FXN and ferredoxin-2 (FDX2), an enzyme intervening at the next step to that catalyzed by FXN. Our data show that these two enzymes compete with each other for binding to the Fe-S cluster biosynthesis complex, thereby repressing each other's activities. Hence, an small excess of FDX2 relative to FXN decreases the efficiency of the reaction, suggesting that a decrease in FDX2 levels could increase the efficiency of Fe-S cluster synthesis in FXN-deficient conditions such as in FA patients. We were able to validate this hypothesis in vivo in a drosophila model of FA, showing that decreasing FDX2 levels improves fly survival. Our data suggest that FDX2 could be used as a novel therapeutic target for FA. In parallel, we have identified compounds by high-throughput screening that stimulate Fe-S cluster biosynthesis and we suspect that they act by alleviating FDX2 repression. The objectives of this project are to elucidate the mode of action of these compounds, better understand the cross-regulation between FXN and FDX2 and test the molecules in the AF drosophila model in order to evaluate their potential as drug candidates. By targeting the primary defect of AF, i.e. the defect in the synthesis of Fe-S clusters, we hope to obtain drugs with high therapeutic potential.
This project is positioned in the CEA's "Biotechnology of Tomorrow" strategic axis, at the interface of life sciences and engineering to address a public health issue. This project combines biology and technologies for the development of a new axis in biotherapy. It is based on a previous CFR thesis funded by the CEA (K. Want 2021-2024), which enabled the development of an anaerobic in vitro screening platform, unique in France, and the identification of active molecules. This thesis also generated the first results showing the existence of cross-regulation between FXN and FDX2, which are the basis of this new project. The continuation of this project thus meets the CEA's objective of building on work already underway, to intensify their developments with the aim of ensuring clinical transfer and scaling up for industrial transfer. Furthermore, this project will rely on the CEA's expertise in modelling and on the I2BC biophysics platforms. It therefore seems important that this project be supported and funded by the CEA.
Miniaturized fluidized bed for the efficient capture of patogens
Sepsis is a generalized, inappropriate immune response to the presence of microorganisms in the blood. This condition is a major public health problem, accounting for over 11 million deaths worldwide every year. One of the reasons for this high mortality rate is the difficulty in rapidly identifying the pathogen involved, thus delaying the rapid administration of appropriate treatment.
This thesis project, in conjunction with the IHU PROMETHEUS, focuses on the development of miniaturized fluidized beds to concentrate blood biomarkers of sepsis, present in trace amounts in biological matrices. This method, based on the use of microfluidic technology, has the potential to replace lengthy blood culture methods, enabling rapid capture and subsequent identification of enriched targets. Thanks to their high flow rates, very large capture surface area and rapid exchange kinetics, the fluidized beds developed for nucleic acid capture will revolutionize the rapid diagnosis of sepsis.
We propose to develop three axes during this thesis project: 1) development and characterization of the miniaturized fluidized bed; 2) analysis of the system's performance with synthetic DNA; 3) validation of the developed system with model biological samples containing bacterial DNA. This DNA analysis system will pave the way for microfluidic analysis of other sepsis biomarkers.
Interface physics of ferroelectric AlBN/Ga2O3 and AlBN/GaN stacks for power electronics
Commercial aviation accounts for about 2.5% total world CO2 emissions (1bT). A true, long-term, clean perspective eliminating a significant part of CO2 emissions is electric. One viable solution could be the hybrid airplane in which gas turbines are used for take-off and landing and in-flight cruising is electrically powered. Such a solution requires high voltage components. Fundamental research is required to optimize materials for integration into electronic components, capable of sustaining these power ratings.
The original idea of the Ferro4Power proposal is to increase the range of applications of Ga2O3 and GaN based devices by introducing a high breakdown, power electronics compatible, ferroelectric layer into the device stack. The up or down polarization state of the ferroelectric layer will provide an electric field capable of modulating the Ga2O3 and GaN valence and conduction bands, and hence the properties of possible devices, such as Schottky diodes (SBD), hybrid depletion mode transistors for Ga2O3 and high frequency HEMTs for GaN. Our hypothesis is to control the electronic bands of Ga2O3 and GaN using an adjacent AlBN.
We will explore the chemistry and electronic structure of AlBN/Ga2O3 and AlBN/GaN interfaces, focusing on the key phenomena of polarization screening, charge trapping/dissipation, internal fields. The project will use advanced photoelectron spectroscopy techniques including synchrotron radiation induced Hard X-ray photoelectron spectroscopy and Photoemission electron microscopy as well as complementary structural analysis including high-resolution electron microscopy, X-ray diffraction and near field microscopy.
The results should therefore be of interest to both physicists studying fundamental aspects of functionality in artificial heterostructures and engineers working in R & D applications of power electronics.
Scientific promise and social circulation of a “molecule of the year”: the case of Reactive Nitrogen
This PhD project proposes an interdisciplinary study of reactive nitrogen (RN), a family of chemical compounds whose uses, representations, and effects intersect scientific, industrial, environmental, health, and political fields. Positioned at the interface between biochemistry and the sociology of science, this work focuses on a central yet still little-discussed object in the social sciences: reactive nitrogen, and more specifically nitric oxide (NO), named molecule of the year by the journal Science in 1992. The discovery of the physiological role of NO led to the Nobel Prize in Medicine in 1998 and to the emergence of a new field of research supported by significant public and private funding, marked by the creation of new scientific societies, conferences and journals, and by the publication of more than 200,000 scientific articles in 30 years. At the crossroads of biomedical promises and scientific controversies, NO crystallizes a tension between, on the one hand, very strong therapeutic hopes and, on the other, major health consequences that are still poorly understood.
The project is structured around two main axes: Axis 1 – Promises and limits of scientific innovation. This axis aims to analyze the research trajectories surrounding NO since the 1990s: what promises were made, in what contexts, and why were some of them not fulfilled? The study will address both scientific and epistemological obstacles as well as institutional or political dimensions (disciplinary fragmentation, funding, research coordination, etc.). Axis 2 – Circulations, appropriations, and narratives. This axis follows NO and reactive nitrogen across different social arenas – laboratories, industry, regulation, politics, civil society – to understand how this biochemical entity is mobilized, defined, valorized, contested. Particular attention will be paid to the contradictory representations that coexist (beneficial molecule / toxic molecule), to problematic or incomplete uses, and to the effects of these circulations on public policy and social uses.
The approach adopted is multidisciplinary and will combine: i) an understanding of the biochemical nature of the object; ii) tools from the sociology of science, the history of technology, and studies of scientific controversies. The data to be collected will be: i) bibliometric data: use of databases such as OPENALEX / WOS; ii) documentary corpus: in-depth analysis of scientific archives, key publications, patents and institutional reports related to NO and RN, and media coverage of NO since the 1980s; iii) semi-structured interviews: a series of interviews will be conducted with various actors who have contributed to the study and use of NO, including: researchers who have worked on biomedical or ecological applications of NO; scientists who are members of the NO Society, a scholarly community dedicated to advancing knowledge on the NO molecule; industrial stakeholders involved in the technological valorization of reactive nitrogen compounds; policy makers and experts who have overseen regulation or public policies concerning NO and RN.
The expected results and contributions are as follows: From a scientific perspective, this research aims to establish the current state of social and scientific representations of reactive nitrogen / NO, to identify points of friction between academic spheres and public, commercial and political arenas, and to propose an analysis of the mechanisms of promise, valorization and misuse of knowledge. The thesis aims to enrich debates on the conditions of circulation of innovations and on the modalities of knowledge production in the life sciences. From a social and political perspective, it will contribute to a better understanding of the health and ecological issues linked to RN, and will formulate recommendations for decision-makers to better articulate expertise, responsibility and public policy.
From the perspective of supervision and research environment of the thesis... this project will be co-supervised by Jérôme Santolini (biochemist, Senior Researcher at CEA – Laboratory for Oxidative Stress and Detoxification), Michel Dubois (sociologist, Senior Researcher at CNRS) and Catherine Guaspare (sociologist, Research Engineer at CNRS). It will be conducted within two research units: CEA – DRF/Joliot: expertise in NO, reactive nitrogen and systemic redox approaches; GEMASS – CNRS-Sorbonne University: sociology of science and technology.
The doctoral student will benefit from a stimulating research environment, combining scientific investigation and critical reflexivity, with a strong interdisciplinary orientation and closely connected to public health and science communication issues
Accelerated high-resolution anatomical MRI at 11.7T using SPARKLING
Magnetic resonance Imaging (MRI) has become the reference neuroimaging technique for probing brain structure and function non-invasively. In particular, anatomical MRI is a gold standard for clinical imaging diagnosis and research, with T1-weighted imaging being the most commonly used sequence. However, the use of this imaging modality is limited by long acquisition times, especially for high resolution anatomical imaging. In this regard, non-Cartesian sampling can accelerate acquisitions through flexible sampling trajectories like SPARKLING, which can efficiently sample k-space and allow efficient and optimal iterative reconstructions with minimal degradation in image quality. In this PhD thesis, the SPARKLING framework which was originally developed for T2*-w imaging will be extended to MPRAGE T1-w imaging, with a goal to accelerate the acquisitions by a factor of 10-15 times, thereby allowing us to reach 1-mm isotropic acquisitions within a minute. Additionally, for extensions of anatomical imaging schemes involving redundant sampling at different inversion times (TI) like MP2RAGE, we propose a novel interleaved under-sampling acquisition and corresponding reconstruction scheme, which minimizes redundancy across different readouts, allowing us to maximally accelerate the acquisition process. In practice, this is achieved through 3D+time extension of the SPARKLING algorithm, that can be combined through the proposed 4D reconstruction scheme. Finally, the thesis will also focus on characterizing the noise profile in k-space for non-Cartesian acquisitions and its effect on the observed resolution in the reconstructed MR images. This will help us build SNR-optimized sampling trajectories, which will be validated against state-of-the-art and clinically utilized protocols (like MP2RAGE) at varying field strengths from 3T to 11.7T. Benchmarking of all the acquisition schemes will be performed through quantitative metrics and also qualitative radiological evaluations, through collaboration of radiologists at NeuroSpin and AP-HP Henri Mondor hospital.
Elucidating and exploiting the biosynthetic pathways of natural products to produce novel pharmacologically relevant molecules
Antimicrobial resistance (AMR) poses a significant global public health threat, necessitating the discovery of new antimicrobials. Natural products (NPs) are important reservoirs for such molecules. Among them, 2,5-diketopiperazines (DKPs) stand out due to their remarkable biological activities. DKP biosynthesis typically involves a core enzyme known as cyclodipeptide synthase (CDPS), which forms a cyclodipeptide scaffold, followed by one or more tailoring enzymes that introduce chemical modifications, leading to more complex DKPs. While the diversity of DKPs obtained is substantial, it remains limited since the initial cyclodipeptide scaffolds are predominantly composed of aromatic and hydrophobic amino acids.
Recently, novel core enzymes termed RCDPSs have been identified, showing no sequence homology to CDPSs. Notably, these RCDPSs utilize aminoacyl-tRNAs as substrates to synthesize cyclodipeptide scaffolds containing arginine.
This project proposes to investigate these RCDPSs, aiming to enable the biosynthesis of diverse DKPs containing arginine and other charged amino acids. The objectives are to establish the natural repertoire of cyclodipeptide scaffolds produced by these enzymes, understand the molecular basis of their substrate specificity, and ultimately perform enzymatic and metabolic engineering to generate a broader diversity of non-natural DKPs with charged amino acids. The project will be carried out using a range of biological (molecular biology, biochemistry, biophysics) and analytical chemistry (LC-MS) methods, with collaborations involving experts in structural biology and synthetic chemistry. If the project's progress allows, a collaboration will be established with an already identified platform to test the biological activity of the generated compounds.
Innovative pharmacological strategy to counter biohazard toxins
THESIS OBJECTIVE. Develop PROTAC molecules for proteasome-mediated degradation of toxins internalized in host cells, and propose drug candidates for in vivo studies at the end of the thesis.
BACKGROUND AND CHALLENGES. Plant and bacterial toxins are among the most toxic natural substances, and are responsible for fatal diseases such as botulism and tetanus. Once the toxin is internalized in the target cells, immunotherapy is ineffective, and there are no curative treatments for these biomolecules. One way of achieving a major breakthrough in the development of medical countermeasures would be to target the toxin directly into the cytoplasm of host cells using PROTAC molecules. PROTACs are heterobifunctional degraders that specifically eliminate targeted proteins by hijacking the cell's ubiquitin-proteasome system. This recent therapeutic strategy represents an attractive technology for new drug discovery.
METHODOLOGY. To carry out this project, the thesis student will carry out in silico screening campaigns to identify ligands for a toxin and improve their affinity. Key validation experiments will require recombinant production of a toxin fragment, and will be carried out in E. coli. From the most promising optimized ligands, targeted libraries of PROTAC molecules directed against the toxin will be synthesized in collaboration with a team of chemists. The student will evaluate the ability of these molecules to interact with and eliminate the internalized toxin in cultured cells using different approaches, in order to propose drug candidates for in vivo studies at the end of the thesis.