Seminars-Gene regulatory network inference from time series: machine learning approaches

EVENT : C3BI Seminars – Seminars-Gene regulatory network inference from time series: machine learning approaches


Speaker : Florence D’Alché-Buc, from Laboratoire de Statistiques et Applications, Télécom ParisTech, Paris Time : 2PM Date :   June 9th, 2016 Location : Room Jules Bordet  – METCHNIKOFF (67) ,Institut Pasteur, Paris


We consider the well known problem of gene regulatory network inference from time series measurements (gene expression, protein concentration) under the angle of Machine Learning. We address two different instances of this problem.

In the first instance we assume no information about the structure and no perturbed data: we develop a nonparametric dynamical model and a learning algorithm to identify the biological dynamical system and its network structure from the data. It gives us the opportunity to present a novel family of vector-valued models called operator-valued kernel based models. These models extends the well know family of scalar-valued kernel to multiple output prediction as well as structured output prediction.

If the results exhibit very good results compared to other methods, it is clear that the bottleneck of these methods is the lack of additional measurements especially, perturbation data that allow to identify the underlying system accurately. We prone the idea that when the goal of experiments is to estimate a model, data have to be chosen consequently. This introduces the second instance.

In this case, we address the design of experiments together with the identification of a parametric model (differential equations) of the network. The idea is to help the biologist to choose the experimental data, especially perturbation data like knock out, knock down, as well as the sampling rate of the measurements with a limited budget for experiments.The experimental design is thought as a one-player game and is solved using a active learning procedure that allows to both optimize the number and the cost of experiments to perform and the network parameters to be identified. On both problems, we show numerical results on the well known benchmarks of the DREAM Challenges. To conclude, new perspectives about dynamical modeling and machine learning and their use in computational biology are drawn.

N.B. The first work is a joint work with Néhémy Lim and Cédric Auliac, the second work is a joint work with Adel Mezine, Véronique Letort and Artemis Llamosi.


Due to security policy in Institut Pasteur, please register before if you plan to come to this meeting

Seminars – Journée i2b2/tranSMART France pour la recherche bio-médicale

EVENT : C3BI Seminars –


Time : 9:00 AM – 6:00 PM Date :  4th of April 2016 & 5th of April 2016

Auditorium Francois Jacob – BIME (26) ,Institut Pasteur, Paris Centre de Recherche des Cordeliers, rue de l’école de médecine, Paris


Les entrepôts de données cliniques et les plateformes de recherche translationnelle prennent une place croissante dans les domaines de la bioinformatique et de l’informatique médicale. Ce workshop a pour objet de faire un point de l’utilisation de ces outils dans la recherche biomédicale en France, et de partager les expériences.

La participation à la journée est gratuite, mais l’inscription avant le 21 mars est obligatoire (https://goo.gl/lRSuWV). La journée aura lieu le 4 avril 2016 à l’Institut Pasteur (Auditorium François Jacob). Le 5 avril 2016 un hackathon est également organisé pour lancer une dynamique de collaboration technique. Le hackathon aura lieu au Centre de Recherche des Cordeliers, rue de l’école de médecine à Paris. Programme disponible ici

C3BI Seminars – Methodological – Biomolecules Random Walks, Heterogeneities and Model Selection: What Information is accessible from experimental Biomolecules Random Walks?

Biomolecules Random Walks, Heterogeneities and Model Selection: What Information is accessible from experimental Biomolecules Random Walk?

Upcoming Events : C3BI Seminars – Methodological – 09/03/2015 at 02:00 pm in Retrovirus room – LWOFF (22)

Date : 09/03/2015 at 02:00 pm Location : Retrovirus room – LWOFF (22) Speakers/Trainers : Jean-Baptiste Masson, Visiting Scientist from Janelia Farm Research Campus, Ashburn, VA, USA For any questions, suggestions (or to volunteer) for future talks/trainings or general feedback please contact us at c3bi-ask@pasteur.fr

Biomolecules Random Walks, Heterogeneities and Model Selection: What Information is accessible from experimental Biomolecules Random Walks?

The development during the last 20 years of single biomolecule tagging allows unprecedented access to single biomolecule dynamics. Trajectories are now space filling in 2D and densities in 3D are rapidly rising. Thus, large amount of random walks are now accessible. These random walks bear information on both the biomolecule dynamics and on the environment properties. One of the key questions is how to exploit these random walks to gain quantitative information on the biological processes taking place and the nature of these random walks. Growing number of data being accessible multiple statistical hypothesis can be tested. Bayesian Inference [1] is a natural tool to handle multiple environment models, large amounts of data and multiple statistical hypothesis. We will discuss the use of Bayesian Inference to analyse single biomolecule trajectories[2–4], show various local models to describe biomolecule dynamics, methods to analyse multi-scale dynamics and describe transitions to anomalous dynamics [3]. We will also comment on out-of-equilibrium dynamics and stochastic integrals dilemma. Furthermore, numerous estimators lack robustness regarding linking mistakes of the tracking algorithm. We will show how to remove tracking by efficiently sampling the biomolecules assignment graph between images. We will comment the various methods to perform these sampling and discuss the inference of multi-scale fields. We will show time evolving maps at the full cell scale using high density tagging. We will discuss toxins receptor interactions with lipid platforms and Glycine Receptors dynamics at the full cell scale in both neurons and transfected Hela Cells. Finally, we will quickly introduce InferenceMAP a user-friendly software to analyse random walks trajectories. [1] U. Von Toussaint. 2011. Bayesian inference in physics. Review of Modern Physics 83:943-999. [2] El Beheiry, M., Dahan, M. and J.-B. Masson. 2015. InferenceMAP: Mapping of Single-Molecule Dynamics with Bayesian Inference. Nature Methods (In Press). [3] Masson, J.-B, Dionne, P., Salvatico, C., Renner, M., Specht, C. G. , Triller, A. and M. Dahan. 2014. Mapping the Energy and Diffusion Landscapes of Membrane Proteins at the Cell Surface Using High-Density Single-Molecule Imaging and Bayesian Inference: Application to the Multiscale Dynamics of Glycine Receptors in the Neuronal Membrane. Biophysical Journal 106 (1), 74–83 . [4] Masson,J.-B., Casanova, D. , Tu ̈rkcan, S., Voisinne, G., Popoff, M. R., Vergassola, M. and A. Alexandrou. 2009. Inferring Maps of Forces Inside Cell Membrane Microdomains. Physical Review Letters 102 (4), 048103.

 

C3BI

C3BI Seminars – !hacking@pasteur – What Health : Bring to the global health the key of its own digital revolution.

@EnguerrandH will present @whathealthfr association @aiderpasteur Thursday 27th of August 2PM in Lwoff building

Upcoming Events : C3BI Seminars – !hacking@pasteur – 08/27/2015 at 02:00 pm in Retrovirus room – Lwoff building

Date : 08/27/2015 at 02:00 pm Location : Retrovirus room – LWOFF (22) Speakers : Enguerrand Habran, Co-founder & President of What Health Association For any questions, suggestions (or to volunteer) for future talks/trainings or general feedback please contact us at c3bi-ask@pasteur.fr

What Health : Bring to the global health the key of its own digital revolution.

What Health is an independent association whose mission is to make emerge and support innovation in healthcare, in particular thanks to the potential of the new technologies. Our process is conceived to take innovation at source and bring the right environment for its growing, to allow its establishment in healthcare system. What Health fosters exchanges and the collaborative working between several universe to help them become players in innovation in healthcare system. What Health approach is diverse: • « Bottom-up », the innovation is coming from the ground and borne either by healthcare practitioners or by patients. • Collaborative, each player of our community that is willing to contribute to improving healthcare system can provide his know-how and expertise. • Global, What Health supports innovation from its emergence to implantation in the healthcare system. • Supported, What Health is partner of major players in the healthcare system. • Ethical, What Health supports all projects that can improve health. The What Health process consists of several actions: • The “Cafés”, a meeting and debate place • The Innovation challenge, collaborative events that allow to prototype a solution and validate the proof of its concept; • The Hub, support structure dedicated to the development and establishment of projects. As a facilitator, What Health makes no claim to intellectual property emerging from events and other activities. That applies also for any actor and partner of What Health. Therefore, the ownership of a project belongs to the team that supports it. Website : http://what-health.org

C3BI

C3BI News – REGATE, REGISTRATION OF GALAXY TOOLS IN ELIXIR

ReGaTE is a software component enabling the automated publication of Galaxy tools into the ELIXIR Tools and Data Services Registry

Upcoming Events : C3BI News – Galaxy Community Conference 2015

Date : 6-8th July 2015 Location : The Sainsbury Laboratory, Norwich, UK

REGATE, REGISTRATION OF GALAXY TOOLS IN ELIXIR

Olivia Doppelt-Azeroual1, Fabien Mareuil1, Eric Deveaud1, Matus Kalas2, Hervé Menager1 1 Center of Bioinformatics, Biostatistics and Integrative Biology, Institut Pasteur, Paris, France 2 Computational Biology Unit, University of Bergen, Norway ReGaTE is a software component enabling the automated publication of Galaxy tools and workflows into the ELIXIR Tools and Data Services Registry (https://elixir-registry.cbs.dtu.dk/#/). This registry is a web portal for the exploration of bioinformatics resources, such as software packages, web services, websites, or reference databases. Through a dedicated interface, its users can search and locate relevant tools and data resources, and bioinformatics resource providers can enhance the visibility of their services. The registration of resources in the registry can be performed either manually, by filling a form on a web user interface, and providing the required description elements, or automatically by using the registry API. ReGaTE uses the BioBlend API and the Registry API to completely automate the registration of the tools installed on any given Galaxy portal. Central to the development of this tool is the mapping of the Galaxy datatype system to the EDAM Ontology. EDAM provides a controlled vocabulary for the description of scientific topics, software operations, types of data and data formats and it is used to describe the contents of the ELIXIR Registry. This mapping enables the automation of the registration of Galaxy tools by describing the format of their input and output data in the controlled vocabulary of the registry. This mapping is being developed in collaboration with members of the Galaxy team, the EDAM ontology and the Common Workflow Language project. ReGaTE is available at http://github.com/bioinfo-center-pasteur-fr/ReGaTE

C3BI

C3BI News – JOBIM 2015

Most of HUB Team members are attending to JOBIM conference this week.

Upcoming Events : C3BI News – JOBIM 2015 – Clermont Ferrand

Date : From Monday 6th July 2015 to Thursday 9th July 2015 Location : Polydome – Clermont-Ferrand

JOBIM 2015

Most of HUB Team members are attending to JOBIM conference this week. You can have a look to our poster n°116 PosterHubC3BI

C3BI

C3BI Seminars – Large audience – Comparative genomics in vertebrates: from models to applications.

Hugues Roest Crollius @ENS_ULM will talk about Comparative genomics in vertebrates: from models to applications. #C3BIPasteur #LargeAudience

Upcoming Events : C3BI Seminars – Large audience – 07/02/2015 at 02:00 pm in Retrovirus room – LWOFF (22)

Date : 07/02/2015 at 02:00 pm Location : Retrovirus room – LWOFF (22) Speakers/Trainers : Hugues Roest Crollius, Head of the Genome Evolution and Dynamics Group (DYOGEN) from Ecole Normale Supérieure For any questions, suggestions (or to volunteer) for future talks/trainings or general feedback please contact us at bioinfo-hub@pasteur.fr

Comparative genomics in vertebrates: from models to applications.

Genomic rearrangements, first observed more than a century ago, are one of the most obvious differences between genomes of closely related eukaryotes. Genomic rearrangements modify the order and organisation of genes because of evolutionary breakpoints caused by chromosome inversions, translocations, fusions or fissions. They do not occur randomly in genomes and are known to be associated with specific genomic features, yet the forces that promote them have remained elusive. Importantly, chromosome breakpoints are mutational events that may cause or may be associated with human pathologies. I will present two facets of comparative genomics in vertebrates that exploit the properties of genome rearrangements. The first illustrates how considering breakpoints in an evolutionary framework can lead to statistical models that formulate strong and plausible explanations for the mechanisms that cause breakpoints. The second postulates that genomic regions that do not rearrange during evolution are likely to be under strong functional constraints, thus revealing potential long-range cis-regulatory domains. Methods to identify such regions were experimentally tested both in model organisms and with respect to their involvement in human developmental disorders. Comparative genomics is a powerful multidisciplinary approach to decipher genome biology, and the recent acceleration in genome sequence production is paving the way for exciting discoveries. Website

C3BI

C3BI Meeting – Seminar – Methodological – Modelling and visualization of high dimensional biomedical data

Finding biologically relevant structure and patterns in high dimensional multivariate data

Upcoming Events : Seminar – Methodological 06/18/2015 at 02:00 pm at Retrovirus room – LWOFF (22)

Date : 06/18/2015 at 02:00 pm Location : Retrovirus room – LWOFF (22) Speakers : Magnus Fontes, Head from International Group for Data Analysis at Institut Pasteur in Paris For any questions, suggestions (or to volunteer) for future talks or general feedback please contact us at bioinfo-hub@pasteur.fr

Modelling and visualization of high dimensional biomedical data

We will look at some different strategies for finding biologically relevant structure and patterns in high dimensional multivariate data. These strategies are all coupled to specific selections of underlying metrics and optimization of corresponding objective concentration functions. Please join us and share your experience, views and thoughts on the talk. We want this meeting to have an interactive and lively atmosphere with many debates.

C3BI

C3BI Meeting – Seminar – Large audience – Combating cancer: from magic bullets to synthetic interactions and signaling network modeling

From magic bullets to synthetic interactions & signaling network modeling talk by Emmanuel Barillot @institut_curie #cancer #C3BIPasteur

Upcoming Events : Seminar – Large audience 06/04/2015 at 02:00 pm at François Jacob Amphitheatre in the François Jacob building (BIME – 26)

Date : 06/04/2015 at 02:00 pm Location : François Jacob Amphitheatre in the François Jacob building (BIME – 26) Speakers : Emmanuel Barillot, Head of Department of Computational Systems Biology of Cancer at Institut Curie, U900 INSERM, Mines ParisTech. For any questions, suggestions (or to volunteer) for future talks or general feedback please contact us at bioinfo-hub@pasteur.fr

Combating cancer: from magic bullets to synthetic interactions and signaling network modeling

Four decades of cancer molecular biology research have led to the identification of many molecular determinants of this pathology, and resulted in significant progress in medical treatment. First concepts like magic bullets and oncogene addictions were introduced with the idea that one single gene might have causal role in cancer, or that a strategy of treatment could be focused on one supposedly single gene fragility point. Though it bore fruits in biology and in clinics, this reasoning has shown its limitations when it appeared that the molecular pathways governing tumorigenesis and tumor progression are tightly interconnected in a complex network of interactions which covers essential cell processes named hallmarks of cancer (DNA repair, proliferation, death, differentiation, immune response…), and whose crosstalks, feedbacks and compensatory mechanisms invalidated the former simplistic reasoning and on the clinical side provided a profusion of tumor escape paths to existing treatments. The concept of synthetic pairwise interactions between genes that jointly control phenotypes was then introduced, and shed light on cancer biology as well as it opened perspectives of new therapeutic strategies, that exploits the availability of a growing list of targeted inhibitors which are able to inactivate specifically one given protein. The principle of synthetic pairwise interaction is very fruitful, but also shows limits, and it appears today that the modeling of tumor signaling network is required to explore these interactions, explain the biology of cancer, and design innovative treatment strategies. The talk will present several steps in this direction. Please join us and share your experience, views and thoughts on the talk. We want this meeting to have an interactive and lively atmosphere with many debates.

C3BI

C3BI Meeting – Seminar – Methodological – Bayesian network approaches to modelling environmental and human health.

Upcoming Events : Seminar – Methodological 06/01/2015 at 02:00 pm at Jean-Paul Aubert room – FERNBACH (68)

Date : 06/01/2015 at 02:00 pm Location : Jean-Paul Aubert room – FERNBACH (68) Speakers : Mark Borsuk from Thayer School of Engineering and Institute for Quantitative Biomedical Sciences (iQBS), Dartmouth College, Hanover, NH, USA (https://engineering.dartmouth.edu/people/faculty/mark-borsuk) For any questions, suggestions (or to volunteer) for future talks or general feedback please contact us at bioinfo-hub@pasteur.fr

Bayesian network approaches to modelling environmental and human health.

Bioinformatics is concerned with the use of advanced computational methods to: (1) further our understanding of biological systems at all levels of detail and (2) support rational and evidence-based decision making concerning problem prevention, identification, and management. One integrative approach for pursuing these goals is Bayesian network (BN) modeling. By succinctly and effectively translating causal assertions into patterns of probabilistic dependence and vice versa, BNs facilitate logical and holistic reasoning under uncertainty in complex systems. Such structured probabilistic reasoning is necessary for accurate learning, analysis, synthesis, prediction, and decision making. I describe the BN approach, with reference to a number of examples from my own research in the areas of environmental and human health. Recent work in my group has focused on using BNs to reveal interactions between genetic and environmental risk factors in causing human disease. We have developed novel algorithms for learning the structure of a BN from a combination of prior knowledge and observational data. By capturing additive and non-additive interactions among multiple genetic and environmental risk factors, BN structures can improve understanding of biological processes as well as provide a tool for individualized risk assessment. Please join us and share your experience, views and thoughts on the talk. We want this meeting to have an interactive and lively atmosphere with many debates.

C3BI