HAL-MINES ParisTech
Not a member yet
27348 research outputs found
Sort by
Apprentissage automatique et biologie des systèmes pour identifier des stratégies thérapeutiques dans le cancer du sein triple négatif atip3-déficient
No full textTriple-Negative Breast Cancers (TNBC) with low expression of the microtubule-stabilising protein ATIP3 (encoded by the MTUS1 gene) represent one of the most aggressive subtypes of breast tumours.This thesis has four complementary objectives:(1) to identify the mechanisms responsible for the aggressiveness of ATIP3-deficient TNBCs (ATIP3-) compared to ATIP3-proficient tumors, (2) to identify therapeutic targets specific to the ATIP3- TNBC subtype, (3) to propose compounds capable of modulating these targets, and (4) to facilitate the identification of such compounds by developing chemogenomic algorithms, which more generally could accelerate drug discovery.First, the construction of a network of 17 microtubule-related genes, including MTUS1, allowed us to highlight functional deregulations in breast cancer at the level of the mitotic spindle, particularly during the G2/M transition. Second, to test the hypothesis that ATIP3 deficiency may propagate deregulations beyond microtubule-related functions, we analysed transcriptomic datasets from ATIP3- breast tumours as well as from tumours of other origins. These analyses identified a small set of functionally interconnected biological pathways deregulated in ATIP3- TNBCs: hyperactivation of the oncogene MYC associated with impaired activation of FOXO3, overexpression of proteins involved in oxidative phosphorylation supporting the metabolism of highly proliferative cells, activation of stress response mechanisms to misfolded proteins in response to proteotoxic overload (UPR pathway), and activation of DNA repair mechanisms in response to replication stress. The cooperation of these mechanisms helps ATIP3- TNBC cells answer the specific needs of highly proliferative cells, contributing to tumour aggressiveness in addition to mitotic spindle deregulations.Using gene dependency scores and large-scale compound response data from TNBC cell lines, we showed that the above pathways (whose activation was required in ATIP3- TNBCs) also included genes/proteins that represented vulnerability points, thus suggesting therapeutic targets. We identified a restricted set of proteins (MYC, PSMB5, TCP1, MASTL, CHEK1/WEE1, AURKA-CDC42, HDAC3, BAG2, DDX10), associated either with drugs already used in the clinic or with molecules still at the experimental stage (alisertib, bortezomib, luminespib, prexasertib, adavosertib). Overall, these results suggest that therapeutic approaches combining the deregulation of the G2/M checkpoint, exacerbation of proteotoxic stress through the inhibition of proteins in the UPR pathway, or targeting proteins involved in DNA repair may be effective.In parallel, this thesis introduces methodological tools for predicting drug-target interactions (DTIs). We propose the Komet model, a fast and scalable chemogenomic algorithm trained on a large dataset assembled during this work (LCIdb). Komet achieves state-of-the-art prediction performance without relying on much more computationally expensive deep learning architectures. We illustrate its utility on the WEE1 target: Komet retrieves known inhibitors and proposes new structurally distinct chemotypes.In conclusion, this thesis combines computational and biological approaches to propose a set of testable therapeutic hypotheses in in vitro models, while providing open and reusable tools for precision oncology beyond TNBC.Les cancers du sein triple négatif (TNBC) sous-exprimant la protéine stabilisatrice des microtubules ATIP3 (codée par le gène MTUS1) représentent l'un des sous-types tumoraux les plus agressifs parmi les cancers du sein.Cette thèse poursuit quatre objectifs complémentaires :1) identifier les mécanismes responsables de l'agressivité des TNBC ATIP3-déficients (ATIP3-) par rapport aux tumeurs ATIP3-proficientes, (2) identifier des cibles thérapeutiques spécifiques au sous-type TNBC ATIP3-, (3) proposer des composés modulant ces cibles, et (4) faciliter leur identification en développant des algorithmes de chémogénomique, ce qui, plus généralement, pourrait faciliter la découverte de médicaments.Dans un premier temps, la construction d'un réseau de 17 gènes liés aux microtubules et comprenant MTUS1 nous a permis de mettre en évidence, dans le cancer du sein, des dérégulations fonctionnelles au niveau du fuseau mitotique, notamment lors de la transition G2/M.Dans un second temps, afin d'évaluer l'hypothèse qu'une déficience en ATIP3 pourrait propager des dérégulations au-delà des fonctions liées aux microtubules, nous avons étudié un ensemble de données transcriptomiques issues de tumeurs du sein déficientes en ATIP3, mais aussi issues de tumeurs d'autres localisations. Ces analyses ont permis d'identifier un petit nombre de voies biologiques fonctionnellement interconnectées, et dérégulées dans les TNBC déficientes en ATIP3: une hyperactivation de l'oncogène MYC associée à un déficit d'activation de FOXO3, une surexpression des protéines impliquées dans la phosphorylation oxydante permettant de soutenir le métabolisme de cellules fortement proliférantes, une activation des mécanismes de réponse au stress de protéines mal repliées en réponse à une surcharge protéotoxique (voie UPR), et une activation des mécanismes de réparation de l'ADN en réponse au stress de réplication. La coopération de ces mécanismes aide les cellules TNBC déficientes en ATIP3 à couvrir les besoins spécifiques des cellules fortement proliférantes, et contribue à l'agressivité de ces tumeurs, en complément des dérégulations du fuseau mitotique.À l'aide de données de scores de dépendance génique et de données à grande échelle de réponse à des composés sur des lignées TNBC, nous avons montré que les voies biologiques ci-dessus (et dont l'activation était requise dans les TNBC déficientes en ATIP3) comprenaient également des gènes/protéines qui constituaient des points de vulnérabilité, suggérant ainsi des cibles thérapeutiques. Nous avons ainsi isolé un ensemble restreint de protéines (MYC, PSMB5, TCP1, MASTL, CHEK1/WEE1, AURKA-CDC42, HDAC3, BAG2, DDX10), associées à des médicaments utilisés en clinique ou à des molécules encore au stade expérimental (alisertib, bortezomib, luminespib, prexasertib, adavosertib). Globalement, ces résultats suggèrent des approches thérapeutiques combinant des dérégulations du checkpoint G2/M, une aggravation du stress protéotoxique par inhibition de protéines de la voie UPR, ou en ciblant des protéines impliquées dans la réparation de l'ADN.Parallèlement, cette thèse propose des outils méthodologiques pour la prédiction d'interactions médicament-cible thérapeutique (DTI). Nous proposons le modèle Komet, un algorithme de chémogénomique rapide et passant à l'échelle, entraîné sur un large jeu de données assemblé au cours de ce travail (LCIdb). Komet obtient des performances de prédiction de pointe, sans recourir à des architectures d'apprentissage profond beaucoup plus coûteuses en temps de calcul. Nous illustrons son utilité sur la cible WEE1 : Komet retrouve des inhibiteurs connus et propose de nouveaux chémotypes structurellement distincts.Au final, cette thèse propose, grâce à des approches computationnelles et biologiques intégrées, un ensemble d'hypothèses thérapeutiques testables dans des modèles in vitro, tout en fournissant des outils ouverts réutilisables en oncologie de précision au-delà des TNBC
Influence of a copper interlayer on the electrical contact resistance of a nickel-silver-plated interface subjected to fretting wear
No full textInternational audienceSensors and electrical connectors are increasingly used in most industries such as mechanical, automotive or aerospace. Most of these mounted systems are subjected to vibrations favoring fretting wear and delamination of protective coatings [1, 2]. The studied electrical contacts consist in a multilayer structure involving a brass substrate coated with an intermediate nickel layer to limit the diffusion of copper from the substrate and finally a top silver plating which is delaying the formation of insulating oxide debris. However, some adherence problems can be observed between the top silver layer and the intermediate nickel layer. One solution to improve this adherence is to apply a thin “strike” copper interlayer between the two nickel and silver thicker coatings. In this study, the influence of the thickness of the copper interlayer (eCu) regarding the fretting electrical contact resistance is investigated. For this purpose, a homogeneous brass/Ni (2 μm)/Cu (Xμm)/Ag (2 μm) crossed cylinder interface is investigated applying representative gross slip fretting slidings. The results show a significant increase in Electrical Contact Resistance endurance when a thin copper interlayer is applied. The electrical life increases from Nc = 105 cycles without copper interlayer up to Nc = 1.8 × 105 cycles when eCu = 0.2 μm. However, this evolution is not monotonous. Above this threshold thickness eCu = 0.2 μm, a critical decrease in Nc is observed. This suggests that the copper interlayer must be thick enough to improve the nickel-silver adhesion, but must remain thin enough to limit the amount of copper embedded below the silver layer which diffusing to the top surface and forming poisoning copper oxides can decay the electrical contact resistance
Sensitivity of intracranial haemodynamics towards varying arterial tree extensions
No full textInternational audienc
Graph neural networks for real-time prediction and treatment decision in intracranial aneurysms
No full textInternational audienc
A Continuum Approach With Adaptive Mesh Refinement for Platelet Plug Formation
No full textInternational audiencePlatelet plug formation is a critical physiological response to vascular injury, serving as a cornerstone of primary hemostasis. Understanding and simulating this process are essential for advancing patient‐specific treatments and interventions. However, achieving a balance between model accuracy and computational efficiency, in particular, for patient‐specific scenarios, remains a challenge. In this work, we present a continuum‐based approach for simulating platelet plug formation using adaptive mesh refinement, providing a novel solution in this field that enables both accuracy and computational feasibility. Indeed, it integrates a stabilized finite element method within the Variational Multiscale framework to model blood flow dynamics, treated as a non‐Newtonian fluid, along with the transport of biochemical species such as platelets and agonists. The platelet plug is represented by an extra stress term in the Navier–Stokes equation, capturing its influence on local blood flow dynamics as a rigid body. A key feature is related to anisotropic mesh adaptation, enabling high‐resolution representation of the evolving platelet plug boundary while drastically reducing computational cost. We validate the model against two‐dimensional benchmarks under varying shear rates and apply it to a 3D scenario, demonstrating its scalability and precision in simulating thrombosis under complex hemodynamic conditions. The results highlight the model's unique capability to facilitate accurate and efficient patient‐specific simulations, offering a transformative tool for advancing personalized medicine
Theoretical Analysis of Topotomography Using Small Intragranular Strain Approximations
No full textTopo-Tomography (TT) is a synchrotron-based X-ray diffraction imaging technique used to characterize grain shape and crystal orientation in polycrystalline samples. This work aims to provide a decisive and fundamental understanding of 3D grain shape and orientation field reconstruction from TT diffraction data. We derive mathematical expressions for the TT projection geometry, considering grain shape, intragranular lattice rotations, and elastic strains, under the assumption of kinematical diffraction. These expressions are simplified using approximations for small strain variations and grain size. The simplified expressions show that integrated TT projection images correspond to projections of a "pseudo" distorted grain volume. Its Fourier analysis provides insights into the feasibility of orientation field reconstruction from TT scans. We propose methods to expand data coverage, including using opposite scattering vectors and varying detector distance. A lower bound for orientation sampling resolution is derived and validated through simulations
The Care of Things: Ethics and Politics of Maintenance
No full textInternational audienceWhat does a coffee machine, a car, road signs, a smartphone, a cathedral, a work of art, a satellite, a bicycle, a washing machine, a bridge, a watch, a computer, the body of a prominent politician and a tractor have in common? Pretty much nothing – except for the fact that, no matter how small, large, important or insignificant something is, it rarely survives without being cared for. Every object eventually experiences wear and tear, it deteriorates, stops working or breaks down. But are we giving the care of things the recognition it deserves? A counterpoint to our modern obsession with innovation but less striking than the one-off act of restoration, the delicate act of making things last rarely attracts our attention.This book disrupts our dominant narratives by putting those individuals skilled in the art of maintenance front and centre. Jérôme Denis and David Pontille shine a spotlight on the subtle aspects of caring for things, tracing the stories of those involved and, with them, the ethical challenges raised and political lessons learned. These people demonstrate a sensitivity and attentiveness to fragility; they encourage us to cultivate a material diplomacy in which wear is accepted and our relation to things becomes a matter of negotiation and compromise – a far cry from the frenetic rhythm of planned obsolescence inherent in hyper-consumerism. Maintenance demarcates the contours of a world in which we have relinquished the human longing for unlimited power and technological autonomy, a world where our attachment to things is more profound than we ever imagined
A Proximal-Type Method for Nonsmooth and Nonconvex Constrained Minimization Problems
No full textInternational audienceAbstract This work proposes an implementable proximal-type method for a broad class of optimization problems involving nonsmooth and nonconvex objective and constraint functions. In contrast to existing methods that rely on an ad hoc model approximating the nonconvex functions, our approach can work with a nonconvex model constructed by the pointwise minimum of finitely many convex models. The latter can be chosen with reasonable flexibility to better fit the underlying functions’ structure. We provide a unifying framework and analysis covering several subclasses of composite optimization problems and show that our method computes points satisfying certain necessary optimality conditions, which we will call model criticality. Depending on the specific model being used, our general concept of criticality boils down to standard necessary optimality conditions. Numerical experiments on some stochastic reliability-based optimization problems illustrate the practical performance of the method
On the Computation of Constrained Wasserstein Barycenters
No full textInternational audienceThis work presents two optimization methods to compute, subject to constraints, a Wasserstein barycenter (WB) of finitely many empirical probability measures. The new measure, denoted by constrained Wasserstein barycenter, extends the applicability of the standard WB to pre-required geometrical or statistical constraints. Our first approach is an extension of the Method of Averaged Marginals (Mimouni et al., 2024) to compute WBs subject to convex constraints. In the nonconvex setting, we propose an optimization model whose necessary optimality conditions are written as a linkage problem with non-elicitable monotonicity. To solve such a linkage problem, we combine the Progressive Decoupling Algorithm (Rockafellar, 2019) with Difference-of-Convex programming techniques. We give the mathematical properties of our approaches and evaluate their numerical performances in two applications, demonstrating both their computational efficiency and the practical relevance of constrained Wasserstein barycenters