Projects

  • Bilateral contracts with industry

    • Ovega

      A contract is signed with the company “Ovega” to exploit the results of the action exploratoire “Community Garden Book

  • International initiatives

    • CNRS IAL PredEvo

      The Beagle team is part of the CNRS International Associated Laboratory “PredEvo”. The two other members of the project are the Beacon Center (Michigan State University, East-Lansing, US) and the TIMC-IMAG (CNRS, Université Grenoble-Alpes, France).

  • National initiatives (ANR)

    • ABC4M, 2020-,

      Approximate Bayesian computation-driven multimodal microscopy to explore the nuclear mobility of transcription factors, a project funded by the French National Agency for Research (ANR), Call “AAP génériques 2020″. We combine computer simulations and Approximate Bayesian computation with simultaneous multiple microscopy methods (FCS and spt-PALM) to quantify the way transcription factors explore the nucleus to find their binding sites. The project is supervised by H. Berry. Other participants are Institut Langevin, ESPCI, Paris (I. Izeddin), Phlam laboratory, Lille (L. Héliot) and Univ. Berlkeley, CA, USA (X. Darzacq). Total amount funded: 565keuro.

    • EngFlea (Engram of fast learning in the striatum), 2021-,

      Call AAPG ANR 2021. Our goal is to study the link between endocannabinoid-dependent plasticity and fast learning of rodents thanks to a multidisciplinary approach combining in vitro and in vivo experimental neurophysiology with detailed subcellular biophysical models and large-scale neural network models. Supervisor: L. Venance (CIRB, Collège de France, Paris). Participant: H. Berry.

    •  Evoluthon (2019-2022)

      Artificial Life as a benchmark for evolutionary studies, a 4-year project leaded by E Tannier with 2 partners, Beagle Inria and Le Cocon, LBBE.

    •  Storiz (2018-2021):

      Horizontal transfers as documents from extinct or unknown species. Call ANR JCJC 2018. Leaded by Damien de Vienne (LBBE, Lyon) Participant: Eric Tannier

    •  LncEvoSys (2018-2021):

      An evolutionary systems approach to understand long non-coding RNA functionality, Call ANR JCJC 2017. Leaded by Anamaria Necsulea (LBBE, Lyon). Participant: Eric Tannier

    •  ANR Equipex+ grant “Spatial Cell Id” (2021-)

      coordinated by Yad Ghavi-Helm (IGFL), Olivier Hamant (RDP), and Jonathan Enriquez (IGFL) – 4,2M€. Anton Crombach and Christophe Godin are  contact persons between Inria teams (Beagle, Dracula, Mosaic) and Yad, Olivier, Jonathan.

    •  NeGA 2021-,

      Ne effect on Genetic Architecture. By studying several eukaryotic species as well as evolution models like Aevol, NeGA aims at a better understanding of the influence of the effective population size (Ne) on the Genetic Architecture of these species. The project is supervised by Tristan Lefebure (LEHNA, Lyon). Other participants are the Beagle team, the LBBE (Lyon) and the ISEM (Montpellier).

  • Inria initiatives

    • Naviscope (Inria Project Lab, 2018-2022): image-guided Navigation and VIsualization of large data sets in live cell imaging and microSCOPy.

      Nowadays, the detection and visualization of important localized events and process in multidimensional and multi-valued images, especially in cell and tissue imaging, is tedious and inefficient. Specialized scientists can miss key events due to complexity of the data and the lack of computer guidance. In Naviscope we develop original and cutting-edge visualization and navigation methods to assist scientists, enabling semi-automatic analysis, manipulation, and investigation of temporal series of multi-valued volumetric  images, with a strong focus on live cell imaging and microscopy application domains. We build Naviscope upon the strength of scientific visualization and machine learning methods in order to provide systems capable to assist the scientist to obtain a better understanding of massive amounts of information. Such systems will be able to recognize and highlight the most informative regions of the dataset by reducing the amount of information displayed and guiding the observer attention. Head: C. Kervrann (Serpico), other EPIs: Aviz, Beagle, Hybrid, Morpheme, Mosaic, Parietal, and MaIage (INRA unit).

    • Action Exploratoire “Community Garden Book”:

      IPBES’s recent report on declining biodiversity calls for generalization of agroecological, productive, biodiversity and environmental friendly methods, oriented towards participatory action research. This exploratory action is a proposal to develop tools from open science, evolution science and algorithmics for the co-construction and use of an agroecological network of interactions between groups, species, varieties found in fields and gardens.

    • Action Exploratoire ExODE:

      In biology, the vast majority of systems can be modeled as ordinary differential equations (ODEs). Modeling more finely biological objects leads to increase the number of equations. Simulating ever larger systems also leads to increasing the number of equations. Therefore, we observe a large increase in the size of the ODE systems to be solved. A major lock is the limitation of ODE numerical resolution so ware (ODE solver) to a few thousand equations due to prohibitive calculation time. The AEx ExODE tackles this lock via 1) the introduction of new numerical methods that will take advantage of the mixed precision that mixes several floating number precisions within numerical methods, 2) the adaptation of these new methods for next generation highly hierarchical and heterogeneous computers composed of a large number of CPUs and GPUs. For the past year, a new approach to Deep Learning has been proposed to replace the Recurrent Neural Network (RNN) with ODE systems. The numerical and parallel methods of ExODE will be evaluated and adapted in this framework in order to improve the performance and accuracy of these new approaches.
      In [24], we propose and validate the great behavior of the induced error of our mixed precision scheme DP-SP (double and simple floating point precision) while preserving the double precision. This method is justified by a mathematical reasoning which affirms its convergence with an average error of the order of e/pN but also verified by various numerical tests which show the compensation of the error with the increase of the system size.
      As we have already seen, this method is characterized by its simplicity, its efficiency and above all its vast field of application, especially in biology with large and complicated systems. By the way, following all these mentioned advantages we note that through this article the study of the precision was done by considering the rounding error, whereas we know well that this is not the only error involved in optimizing accuracy.
      This encourages us to deal with approximation errors, in order to obtain a solver and a numerical scheme compatible with our mixed precision method, so we can be able to offer an optimal precision for large scale systems in future works. In order to do so, we will use existing tools (PROMISE [16] and VerifTracer [6]) to evaluate the numerical quality of our code and quantify the magnitude of floating point related errors. Nonetheless, one of our goal is to improve performance (execution time) of ODE solver. Thus we will do a thorough performance evaluation of our method on the different proposed biologicalsystems. To conclude, we will assess how our method  an benefit from next generation computing platform. Especially, we will work on porting our method to take into account silicon based mixed precision implementations that were tailored for IA/ML.

  • Other national initiatives

    • ARC CEDRiC

      Fondation ARC funds the project CEDRiC, a collaboration of Anton Crombach with Sandra Ortiz- Cuaran (head), Pierre Martinez,  Marene Mahtouk, and Janice Kielbassa from the Cancer Research Center of Lyon (CRCL) / Centre Léon Bérard (CLB). This is a two year grant of 50k€ for experiments (2021-2023)