Présentation

Notre recherche se concentre sur le décryptage des déterminants moléculaires du cancer, avec pour objectif de fournir les connaissances nécessaires pour améliorer la prise en charge des patients. Elle repose sur des analyses de données multi-omiques de haute dimension, caractérisant les tumeurs à différents niveaux, ainsi que sur des approches innovantes d'intelligence artificielle et de modélisation de réseaux biologiques. L'institut Curie offre un cadre idéal pour atteindre ces objectifs, grâce aux nombreuses collaborations possibles avec des biologistes et cliniciens, ainsi qu'à l'accès à un plateau technique de pointe. Nous sommes également engagés dans de nombreuses collaborations nationales et internationales, notamment avec des laboratoires en Espagne, Allemagne, Italie, Norvège, Pays-Bas, États-Unis et Japon. De plus, nous développons des logiciels disponibles publiquement sur le répertoire GitHub de notre équipe : https://github.com/sysbio-curie.
Membres
Publications clés
npj Systems Biology and Applications - 30/10/2023
Voir les auteurs
Bioinformatics - 01/06/2023
Voir les auteurs
Bioinformatics - 13/05/2022
Voir les auteurs
Computational and Structural Biotechnology Journal - 01/01/2022
Voir les auteurs
PLOS Computational Biology - 28/01/2021
Voir les auteurs
Scientific Data - 01/12/2020
Voir les auteurs
Entropy - 04/03/2020
Voir les auteurs
Nature Communications - 01/12/2019
Voir les auteurs
Bioinformatics - 01/04/2019
Voir les auteurs
Nature Genetics - 01/09/2017
Voir les auteurs
Vie d'équipe
We invite applications for a staff research position at the interface of artificial intelligence and computational biology of cancer. The position focuses on development and application of machine learning approaches, in particular large language models and transformer-based architectures, to advance our understanding of complex biological systems.
More information on the job offer.
We are looking for a PhD candidate ! During this thesis, the candidate will extend PhysiCell, a multiscale simulation framework to model in vitro cancer systems, with a primary focus on glioblastoma (GBM) microenvironmental dynamics.
The thesis will take place within the framework of PR[AI]RIE-PSAI, supervised by Laurence Calzone.
To apply: send CV, motivation letter, copy of last diplomas, and contact details of 2-3 references to laurence.calzone - at - curie.fr and indicate as subject the reference PR[AI]RIE-PhD. The deadline to apply is May May 22. The results will be communicated in two phases
before June.
More information here : Description du projet de thèse
Building Multiscale Models to investigate CAF-mediated T Cell
exclusion from tumour nests in Non Small Cell Lung Cancer.
We are looking for a PhD candidate !
The thesis will take place within the framework of PR[AI]RIE-PSAI, co-supervised by Laurence
Calzone and Vincent Noël.
To apply: send CV, motivation letter, copy of last diplomas, and contact details of 2-3 references
to laurence.calzone - at - curie.fr and indicate as subject the reference PR[AI]RIE-PSAI-PhD. The
deadline to apply is May 15. The results will be communicated in two phases between May 30 and
June 15.
More information here : Description du projet de thèse
Computational Computational Systems Biology for Complex Human Disease - 21 avril 2024
Computational Computational Systems Biology for Complex Human Disease: from static to dynamic representations of disease mechanisms
This exciting course will focus on applied, functional analysis and interpretation of disease data using computational modelling tools.
The week-long programme will cover the construction and analysis of both static and dynamic mechanistic networks of disease mechanisms concerning complex human diseases such as cancers, autoimmune, inflammatory diseases and others.
This course is aimed at PhD students, postdocs and clinicians/healthcare professionals who are interested in using systems biology approaches and computational modelling to tackle biological and biomedical problems concerning human disease.
More information : https://coursesandconferences.wellcomeconnectingscience.org/event/computational-systems-biology-for-complex-human-disease-from-static-to-dynamic-representations-of-disease-mechanisms-20240421/
EMBO Workshop - 26 novembre 2023
Computational models of life: From molecular biology to digital twins
Single-cell and spatial omics have increased the data granularity available for understanding biological mechanisms as well as exponentially increasing the amount of biomedical data available. Modelling has helped biomedical projects by uncovering mechanisms behind drug synergies and personalizing drug treatments while allowing for the study of all scales, from molecules to cells to tissues to organs and even to whole bodies. High-Performance Computing (HPC) environments have enabled the scaling up of scientific computing allowing it to reach an unprecedented scope. HPC has become an essential tool in many fields due to its ability to power up data analysis, storage and large-scale processing.
This in-person 5-day EMBO Workshop will gather early-career researchers and field leaders at the crossroads of single-cell omics, modelling and HPC to pave the way for realistic, scalable human digital twins of cells, tissues and organs. Presentations and discussions around these rapidly-evolving topics will push the field forward by training researchers with techniques from neighbouring fields and allowing them to tackle problems using combined approaches.
More information : https://meetings.embo.org/event/23-comp-models-life
Models of data, data for models - 25 septembre 2023
International Course on Computational Systems Biology of Cancer
The diversity across tumors from different patients and even across cancer cells from the same patient makes the picture very complex. The idea of ‘personalized’ or ‘precision’ medicine has been suggested, aiming to find tailored treatment regimen for each patient according to the individual genetic background and tumor molecular profile. This attempt is achievable thanks to sufficient molecular characterization of cancers accumulated using high-throughput technologies and advanced imaging technologies. However, despite availability of cancer multi-scale data, they are not fully exploited to provide the clue on deregulated mechanisms that would guide better patients stratification and to specific treatment in cancer.
The objective of the course is to promote better use of computational approaches into biological labs and to clinics. We aim to help participants to improve interpretation of various types of data accumulated in the labs using multi-modal data integration approaches.
More Info : https://training.institut-curie.org/courses/sysbiocancer2023
Workshop at BC2 conference in Basel - 11 septembre 2023
Mechanistic and AI digital twins in personalized medicine - two sides of the same coin
Digital twins are an emergent concept in Computational Systems Biology and personalized medicine. Many initiatives manifest an interest of creating computational models of increasing complexity that could be used to represent virtual patients and help in the decision-making regarding the appropriate treatment. Having robust and reliable computational models spanning all biological layers such as gene expression, signalling and metabolism to name a few, could revolutionize the way we treat Big Data for the benefit of precision medicine and improved medical care - tailored to the needs of each patient. As the number of computational models rapidly increases, the production of data is ever growing and the approaches, both mechanistic and AI-based, are rapidly developing, discussions about challenges and best practices are needed more than ever.
The purpose of the workshop is to bring together researchers working in the field of digital twins in computational systems biology who use various formalisms to address challenges of data integration and model personalization. The focus will be on presenting the state of the art in the field and how mechanistic computational modelling can upscale benefiting from AI-based methodological advances.
More information : https://bc2.ch/tutorials-workshops
From pathway modelling tools to cell-level simulations - 25 juin 2023
PerMedCoE Summer School 2023
The PerMedCoE summer school will take place on the 25 – 30 June 2023 at the Hotel Montanyà in Seva, less than an hour away from Barcelona.
The course will combine lectures and hands-on exercises to:
- Introduce attendees to key practicalities of working with high-performance computing (HPC) clusters
- Build first-hand experience in using PerMedCoE modelling tools (CellNOpt, CARNIVAL, COBREXA, MaBoSS, PhysiCell and PhysiBoSS)
- Develop a practical appreciation of using workflow managers and containerised PerMedCoE modelling tools to execute biomedical workflows
You can access the material we presented there on our sysbio GitHub: https://github.com/sysbio-curie/Curie-PerMedCoE-Summer-School-2023
Nous rejoindre
Rejoignez une équipe dynamique et passionnée pour faire avancer la recherche.
Il n'y a pas d'offre d'emploi disponible pour cette équipe actuellement.
Voir les autres offres d'emploi du Centre de RechercheMethodes intégratives d'apprentissage
IMMUcan: an integrated European immuno-oncology profiling platform
Immune checkpoint inhibitors have shown promising results in some cancer types, but biomarkers to select patients that might respond to the treatment are still missing. Similarly, rationales for combining immune checkpoint inhibitors with conventional therapy or targeted agents are lacking. IMMUcan project aims at improving this situation.
The goal of IMMUcan is to understand, how the immune system and tumours interact, and the impact of therapeutic interventions
IMMUcan project (https://immucan.eu/) receives support from the European Union (grant 821558) within the Innovative Medicine Initiative 2 program and from the European Federation of Pharmaceutical Industries. IMMUcan was started in March 2019 and will end in August 2024.
The IMMUcan consortium is composed by 17 academic organisations from 8 European countries, 1 patient organisation, 1 SME and 9 pharmaceutical companies. IMMUcan is under the leadership of Merck and EORTC.
IMMUcan proposes an inclusive and integrated European immuno-oncology platform. IMMUcan will access high-quality human biological material (tissue, blood, stool and saliva) and clinical data from patients with colorectal, lung, head & neck, breast, gastric cancers and immune checkpoints inhibitors failures. We have assembled a strong consortium with ten expert clinical centers, access to large volumes of the required tumor types. IMMUcan will mobilize the majority of academic trials running or expected to start recruiting patients during the project period. A centralized workflow for samples, via a state-of-the-art biobank will increase reproducibility as all tissues will be processed and stored in a uniform way, following proofed SOPs. The project will perform in depth immunoprofiling with cutting edge technologies including CyTOFF, single cell RNA-seq, peptidogenomics or microbiome analysis. IMMUcan will analyze the data in order to understand the host/tumour interaction in the absence of treatment (naive population) and with treatment (patients in follow-up) to identify potential predictive biomarker for response to immunotherapy, or develop rationale for combination therapies. IMMUcan will provide an IT platform and legal/ethical contractual framework where participants from both academia and industry can pursue their own independent investigations utilizing the IMMUcan data and effectively testing and improving the functioning and relevance of the database. A sustainability plan will be developed to ensure the collection of follow-up data and the accessibility of the data platform. The use of the data generating throughout this project, for future research, will be supported.
The IMMUCan project workflow

TIPIT: Towards an Integrative approach for Precision ImmunoTherapy in Lung Cancer

Non-small cell lung cancer (NSCLC) is the first cause of cancer-related death in France. NSCLC is diagnosed at a metastatic stage in about 70% of patients.. In this situation chemotherapy combined with pembrolizumab is becoming the preferred option even in patients with a tumor with PD-L1 expression ≥50%. The survival of patients with metastatic NSCLC has been increasing with such strategies, but only for 45-50% of those patients a response can be demonstrated. Therapeutic decision is thus suboptimal, and there is a critical need for biomarkers for response prediction.
Our goal is to optimize our current standard-of-care of immunotherapy combined with chemotherapy in NSCLC through the integration of comprehensive, multimodal assessment of factors predicting the efficacy, including clinical data, pathological, radiological and metabolic imaging data, multi-omics profiling, and immunomonitoring. Our strategy will be based on machine learning techniques for extracting relevant features from each data modality : molecular, cellular and tissular level. Our assumption is that this multi-level approach will combine the advantages and information of each level in an optimal predictor.
Our objectives are then 1) to identify signatures predicting response and long-term survival to combination of immunotherapy (based on pembrolizumab as per current approval in France) plus chemotherapy; 2) to decipher biological pathways and mechanisms modulating immune responses, so as to develop prediction models of patient responses treatment.
PanNet: Predictive models of therapy response in pancreatic neuroendocrine tumors
Pancreatic neuroendocrine tumors (PanNETs) are rare neoplasms of the pancreas and have a mortality rate of 60%. Systemic therapies form the mainstay of treatment, but no molecular markers are currently established that predict an individual patient’s response to any given therapy. Different regimens are thus often administered sequentially until objective response is achieved, placing considerable burden on patients and diminishing therapeutic windows.
The clinical partners in this consortium have access to large clinically and molecularly well-defined cohorts of PanNET patients, including samples and data from the most comprehensive molecular analysis of PanNET tumors to date conducted within the International Cancer Genome Consortium (ICGC). Complementary to this, the participating computational biology groups contribute expertise and computational tools for statistical analysis of high-throughput biological data, integration of heterogeneous biological data, and network modeling. Creating unique synergies, this collaborative project will perform systems biology-based analyses of the PanNET data as well as additional publicly available data sets in order to i) identify novel marker signatures in PanNET samples predicting individual patient response to approved therapies; ii) develop new computational methods and construct models of regulatory networks in PanNET to identify novel opportunities for therapeutic intervention with substances used in treatment of other cancers; and iii) validate predictions derived from these analyses in in vitro and in vivo models of PanNET.
We expect this project to i) Implement precision biomarker models for a better therapeutic stratification of PanNET patients, ii) improve clinical decision making and medical treatment of PanNET patients, iii) improve our knowledge of fundamental biological mechanisms active in PanNET and iv) generate significant methodological advancements in the field of precision medicine and systems biology.
ModPhosphoNet
La modélisation de réseaux permet d'étudier les différents processus cellulaires et leur dysfonctionnement lors de maladies telles que le cancer. L'analyse de grands réseaux reste un défi. Notamment, peu d'études intégratives incluent des données de phosphoprotéomique, une source importante d'informations pour étudier les voies de signalisation, en raison des difficultés liées à l'analyse des données de phosphoprotéome. Dans le médulloblastome (MB), la tumeur cérébrale pédiatrique maligne la plus courante, nous avons pu mettre en évidence que des voies de signalisation importantes sont contrôlées au niveau post-transcriptionnel dans deux sous-groupes parmi les quatre idéntifiés. Ainsi, nous proposons ici une nouvelle stratégie d'intégration et de modélisation pour considérer avec précision les informations sur le phosphoprotéome et ses conséquences fonctionnelles potentielles. Deux principales approches informatiques seront explorées: (i) le modèle à blocs stochastiques (SBM) de réseaux multipartites et (ii) la simulation dynamique de réseaux multi-couches. Nous proposons d'utiliser des algorithmes SBM afin d'identifier des groupes de patients en fonction de leur connexion dans plusieurs réseaux moléculaires (protéomique, phosphoprotéomique et transcriptomique). Les résultats des analyses SBM seront interprétés comme des signatures pondérées par sous-groupe de MB. Basés sur ces premiers résultats, nous construirons des modèles mathématiques logiques spécifiques à chaque sous-groupe. Des simulations stochastiques seront effectuées sur ces modèles pour prédire les points d'intervention possibles. Les cibles prédites seront validées fonctionnellement et leur pertinence thérapeutique évaluée à la lumière des données cliniques. L'étude proposée vise non seulement à réaliser d'importants progrès méthodologiques pour résoudre le problème de la réduction et de la modélisation des réseaux biologiques mais aussi à améliorer la prise en charge thérapeutique dans le MB.
Modélisation mécaniste
PerMedCoE: Exascale-ready cell-level simulations for European Personalised Medicine
Personalised medicine opens unexplored frontiers to treating diseases at the individual level combining clinical and omics information. However, the performance of the current simulation software is still insufficient to tackle medical problems such as tumour evolution and response to treatment at the single-cell level or patient-specific treatments. For this reason, this European centre of excellence pursues the following main objectives:
- To optimise key software for cell-level simulations to the new pre-exascale platforms to contribute to the European Personalised Medicine Roadmap.
- To integrate personalised medicine into the new European HPC/Exascale ecosystem, by offering access to HPC/Exascale-adapted and optimised software.
- To design and complete a comprehensive set of personalised medicine use cases to drive the development of cell-level simulations.
- To build the basis for its sustainability by coordinating personalised medicine and HPC communities and reaching out to industrial and academic end-users with use cases, training, expertise and best practices.
The centre will become the entry point to Exascale-ready cell-level simulation software, able to transform personal omics data into actionable mechanistic models of medical relevance, supporting developers and end-users with know-how and best practices. It will also connect simulation software developers with HPC, HTC and HPDA experts at the CoEs POP and HiDALGO, and work with other biomedical consortia such as ELIXIR and LifeTime, also connecting pre- exascale infrastructures hosted by supercomputing centres such as the BSC and CSC.
Further information: www.permedcoe.eu
Modelling of Immunogenic Cell Death (ModICeD)
La mort immunogénique est un processus cellulaire par lequel les cellules mourantes induisent le système immunitaire. Une série de réactions, depuis les premiers signaux de danger à l’activation de processus immunitaires, active un mécanisme de rétroaction positive où la réponse immunitaire anti-tumorale améliore grandement l’efficacité du traitement qui attaque les cellules tumorales qui ont été épargnées par la chimiothérapie. Pourtant, certains agents cytotoxiques communément utilisés ne sont pas capables d’activer certains aspects de la mort immunogénique. Pour approfondir la compréhension globale de ce processus, nous proposons une approche de biologie systémique incluant des analyses omiques et de la modélisation dynamique.
Le projet développera trois aspects : (1) la construction d’un modèle mathématique décrivant la dynamique intracellulaire de différents types cellulaires et leurs interactions (CD4+, CD8+, cellules dendritiques et cancéreuses) qui interviennent dans la mort immunogénique; (2) la validation expérimentale qui se fera d’abord sur un modèle murin du fibrosarcome dans le but de calibrer le modèle à des données réelles, puis sur des lignées cellulaires et des modèles murins du sein pour valider les prédictions du modèle ; et (3) le développement d’un outil de modélisation de dynamique de populations de cellules de différents types basé sur un formalisme Booléen et exécutant des simulations stochastiques à temps continu.
Le projet permettra de définir les perturbations qui amélioreront la mort immunogénique dans le cadre de thérapies anti-cancéreuses
OncoLogics
Le cancer est la deuxième cause de décès dans l'UE. Plus de 150 000 personnes dans l'UE-28 meurent chaque année d'un cancer colorectal (plus de 10 % de tous les décès liés au cancer). Lorsque la maladie est à un stade avancé et que la chirurgie n'est pas possible, on a recours à une thérapie systémique. Il y a quelques décennies, la chimiothérapie était la seule option, avec une survie globale d'environ un an. Grâce à la chimiothérapie, aux thérapies ciblées et aux immunothérapies, la survie est aujourd'hui d'environ trois ans. Bien que les nouvelles thérapies ciblées offrent de grandes possibilités, le défi consiste à associer ces thérapies aux patients qui y répondront le mieux. La survie à cinq ans est encore bien inférieure à 20 %, ce qui indique clairement la nécessité d'améliorer les outils de stratification des patients et les thérapies personnalisées.
Notre approche de la médecine systémique utilise des modèles informatiques pour la conception de thérapies personnalisées. Les modèles informatiques booléens qui représentent les tumeurs de chaque patient seront utilisés pour prédire leur réponse aux traitements médicamenteux, et les prédictions in silico seront comparées aux données cliniques disponibles sur les patients atteints de cancer et aux réponses aux médicaments dans les cultures de sphéroïdes et d'organoïdes dérivées de patients. Les divergences entre les observations et les prédictions seront analysées pour comprendre pourquoi certains modèles échouent, et grâce à des expériences ciblées, nous améliorerons les modèles booléens qui représentent mieux les patients individuels. Notre plateforme de modélisation améliorée utilisera les données tumorales des patients à partir de matériel cultivé ex vivo, produira une liste restreinte de thérapies prometteuses qui seront ensuite testées sur le matériel cultivé ex vivo, afin de fournir au clinicien des suggestions de thérapie spécifiques au patient.
Nous évaluerons les aspects éthiques de la manière dont une plateforme d'aide à la décision basée sur un modèle peut affecter les médecins, les patients et notre modèle de soins de santé. Nous pensons que notre plateforme d'aide à la décision améliorera les diagnostics, les pronostics et la conception des thérapies pour les cancers à un stade avancé, améliorera le pronostic pour les cancers à un stade avancé et améliorera la stratification des patients pour les essais cliniques.
Anciens projets
IPaediatricCure
Cancer in children is rare but when it happens, clinically prescribed treatment options are not always as efficient as one would hope. Of those that are cured, a substantial proportion suffer long-term serious health consequences from the intensive treatments that are currently required. A major reason for cancer being so difficult to treat effectively is that cancer cells undergo many random changes, which means that each cancer has an essentially unique combination of molecular characteristics. To address this problem, it is important to develop ways of specifically tailoring treatment combinations for the molecular profile of each individual cancer, to maximize cures and to minimize short- and long-term treatment side-effects.
IPaediatricCure (https://ipc-project.eu/) is a project supported by EU Horizon-2020 funding program.
The goal of the iPC project is to collect, standardize and harmonize existing clinical knowledge and medical data and, with the help of artificial intelligence, create treatment models for patients. Armed with these treatment models, scientists will then test them on virtual patients to evaluate treatment efficacy and toxicity, thus improving both patient survival and their quality of life. To accomplish our goals, we have assembled an interdisciplinary team consisting of basic, translational, and clinical researchers—all amongst the leaders in their respective fields—and established strong relationships with European Centres of Excellence, patient organizations, and clinical trials focus on personalized medicine for our proposed case studies. In summary, iPC will address the critical need for personalized medicine for children with cancer, contribute to the digitalization of clinical workflows, and enable the Digital Single Market of the EU data infrastructure.















