Search

International audience; In the current world context, with a globalized production system and highly interconnected farms, understanding and controlling livestock diseases and their spreading are critical issues both for food industry and for public decision-makers. In order to prevent outbreaks, it appears essential to characterize and predict animal trade movements, as major pathogen-inducing pathways between farms. A convenient way to grasp and explore the general structure of this complex system is to use a dynamic graph representation. Indeed, the vertices of this network are the farms (and also the commercial operators such as markets and assembly centers), while the edges represent animal exchanges between farms. This graph is directed (sellers send cattle to buyers), weighted (each edge is labelled with the number of exchanged animals), and time-varying (a transaction occurs at a given time-instant). Furthermore, additional information on the vertices, such as geographical situation, type of farm, ... is usually available. With this representation, mathematical models integrating the dynamics of the network, but also the epidemics spreading on the temporal network and the farmers’ behaviour with respect to trading may be developed and investigated. However, one of the limits of these models is their scalability, whereas cattle-trading networks are usually of substantial size. For instance, the data at our disposal, concerning the exchanges involving French farms from 2005 until 2009 with a daily resolution level, contains millions of edges and hundreds of thousands of nodes. In this context, it is important to build a reduced version of this network, by identifying groups of vertices with common features (in a broad sense). In this paper, we propose an exploratory study of the French cattle-trading network using selforganizing maps and, more particularly, the recently developed R-package SOMbrero, which implements both numerical and relational versions of the algorithm. The output consists in a clustering of the vertices and a nonlinear mapping of the graph, which is a reduced version of the original network. The SOM algorithm may be trained either on vector data, which may be numerical features extracted from the dynamical graph or from snapshots of it, or on relational data, which may be any kernel or dissimilarity computed on the vertices or on the edges, such as, for example, the dynamic shortest-path distance. We will illustrate both versions of the algorithm by providing some of the outputs (see Figures 1 and 2 for an example of the algorithm trained on numerical features extracted from one yearly snapshot of the network) and explaining the benefit of each of them in the process of understanding the inner mechanisms of the network

Ce rapport décrit un modèle de simulation stochastique représentant la dynamique des populations de truies et de porcs dans un troupeau naisseur engraisseur et la transmission indirecte des Salmonelles au sein de ce troupeau. Ce modèle a été initialement développé pendant la thèse de doctorat d'Amandine Lurette Il a ensuite a été repris et étandu dans le cadre du stage de Justine Guillaumont, pour intégrer les différentes conduites en bandes rencontrées en Bretagne. L'effet du mélange de bandes et du type de conduite ont été évalués à partir de ce modèle., This report describes a stochastic simulation model which represents both the pig and sow population dynamics in a farrow-to- finish herd and the indirect Salmonella transmission within this herd. This model was initially developed during Amandine Lurette's PhD thesis . It was then picked up and extended as part of Justine Guillaumont's traineeship to integrate the various batch management systems encountered in Brittany. The effect of batch mixing and management systems were assessed using this model.

PRRSv is responsible for significant worldwide production losses and its control is a major challenge for the swine industry. Vaccination, the main control measure, does not allow to eradicate the infection and only confers a partial protection to the host. This lack of efficiency is mainly due to the strong variability in PRRSv strain virulence, which induces highly variable within-host dynamics. Consequently, there is a real need to better understand the interactions between the virus and the immune response in order to improve PRRSv control. To tackle this issue, a dynamic and deterministic modelling approach was chosen. We developed an original immunological model consisting in an integrative representation of the within-host dynamics. It describes the immune mechanisms at the between-cell scale, including the innate response, the activation and orientation of the adaptive response and their complex regulations by the major cytokines. By the definition of parameter sets based on published experimental and modelling studies, this approach allowed us to explore various scenarios in terms of host susceptibility, strain virulence, as well as dose and duration of viral exposure. Our first results show that similar infection durations associated with contrasted immune dynamics can be explained by the consideration of the immune mechanisms affected by the strain virulence. They provide new insights to explain apparent inconsistencies between experimental data. We then showed that the exposure, whose effect is often neglected, has an impact on the within-host dynamics, which varies depending on the virulence level. Finally, the within-host dynamics induced by the infection of a vaccinated pig is currently under exploration, opening up promising leads to improve vaccine efficiency. All these results provide new insights to guide further experimental and modelling approaches and they offer promising prospects to improve PRRSv control at the herd level by an immuno-epidemiological modelling approach.

The immune mechanisms which determine the infection severity and duration induced by pathogens targeting pulmonary macrophages are poorly known. To explore the impact of such pathogens, it is indispensable to integrate the various immune mechanisms and to take into account the variability in pathogen virulence, host susceptibility, and host exposure to the pathogen. In this context, we developed an original ODE model representing the infection and immune dynamics induced by such a pathogen. Compared to previous modelling studies, we detailed the macrophage-pathogen interactions, the innate immune response, and the cytokine regulations. The adaptive immune response included the main functions of the cellular, humoral, and regulatory orientations. The model obtained has 14 state variables: the pathogen; four effectors of the innate response, consisting of three macrophage states (susceptible, phagocyting, and infected) and the natural killers; three effectors of the adaptive response, representing the cellular, humoral and regulatory responses; seven cytokine groups, consisting of the major pro-inflammatory, the innate antiviral and the immuno-regulatory (IFNg, IL12, IL10, TGFb) cytokines. The main processes integrated in the model are: the pathogen phagocytosis by the macrophages; the macrophage infection; the pathogen excretion by infected macrophages; the recruitment and decay/migration of the macrophages; the activation and decay/migration of the other effectors; the cytokine productions by the immune cells and their decay; the cytokine regulations.We calibrated our model for the Porcine Respiratory and Reproductive Syndrome virus (PRRSv), a major concern for the swine industry. We extracted value ranges for the model parameters from modelling and experimental studies on respiratory pathogens. A sensitivity analysis was used to identify the most influential parameters and to define a realistic reference scenario.We first used our model to explore the influence of strain virulence and host susceptibility on the infection duration and immune dynamics. We obtained contrasted results, suggesting hypotheses to explain the apparent contradictions between published results: high levels of antiviral cytokines and a dominant cellular response were associated with either short, the usual assumption, or long infection durations. In addition, we extracted some synthetic and original elements from our work to characterise immune mechanisms and their impact on the infection duration. We then used our model to explore the impact of host exposure on the infection duration and severity for various levels of strain virulence. We tested several exposure functions to account for experimental inoculations or natural infections. We found that: (i) high exposures induced high viral peaks; (ii) prolonged exposures and high virulences induced prolonged infections; and (iii) the viral peak determined the immune response activation and the adaptive response orientation, whereas the virulence determined the relative levels of adaptive antiviral and immuno-modulatory cytokines (IL10, TGFb).In conclusion, this integrative model provides a powerful framework to go beyond experimental constraints. It could be used to help designing efficient vaccination strategies. It could also provide a base for an immuno-epidemiological model, by identifying the key mechanisms (exposure x immune response) that are responsible for the infection severity and duration.Anglais

absent

FILTREX : A new software for parametric identification and optimal sampling os experiments for complex microbiological dynamic systems by nonlinear filtering. Sofware day at the 8th International Conference Predictive modelling in food

International audience; Porcine Respiratory and Reproductive Syndrome Virus (PRRSV) infection is a major concern for swine industry. Our partial understanding of the interactions between the virus and the immune system is the major reason for the lack of efficient control measures. The PRRS virus replicates mainly in the pulmonary macrophages which: (1) are responsible for inflammation and viral destruction by phagocytosis; and (2) participate in the induction and orientation of the adaptive immune response. Consequently, macrophage infection hampers the whole immune response. The interactions between macrophages and virus during the first steps of infection have not been thoroughly investigated and their influence on the infection resolution is unknown. Here, we propose an original model simulating immune and infection dynamics to explore these complex mechanisms and test biological hypotheses. We highly detail the immune functions and infectious statuses of macrophages and take into account interactions between innate and adaptive responses and cytokines regulations. We use the model to study the relative influence of macrophage – virus interactions on the infection resolution comparatively to adaptive mechanisms.

Poster; International audience; PRRSV replicates mainly in the pulmonary macrophages which (i) are responsible for inflammation and viral destruction by phagocytosis and (ii) participate in the induction and orientation of the adaptive immune response. Experimental studies have shown that PRRSV is able to inhibit innate immunity, to reduce the gamma interferon synthesis and to limit the efficiency of neutralizing antibodies. Moreover the virulence and the resulting interaction with the host immune system are variable between viral strains. Our limited understanding of the interaction between the virus and the immune system is the main obstacle in the evaluation of control measures and the development of more efficient vaccines. To explore these complex mechanisms and test biological hypotheses, we propose an original model of the immune response centred on macrophage - virus interactions in the lung. Comparatively to previous modelling studies we highly detailed the temporal dynamics of the innate immune response to better understand the effect of the macrophage-virus interactions on the viral clearance. To explore the relative influence of the immune mechanisms on the infection outcome we conducted a multivariate sensitivity analysis using the R package multisensi. We showed that the first steps of macrophage-virus interactions are crucial. We used the model to explore the immune response to strains of variable virulence and to determine the required conditions for viral clearance.

International audience; Respiratory viruses are responsible for tissue damages and local inflammation. The best strategy to control their severity is to limit the infection while maintaining an efficient immune response. Given this context, the case when the macrophage is the target cell of infection is of interest. Indeed, pulmonary macrophages (i) are responsible for inflammation and viral destruction by phagocytosis and (ii) participate in the induction and orientation of the adaptive immune response. Consequently, macrophage infection hampers the whole immune response. The interaction between macrophages and virus during the first steps of infection has not been throughly investigated in experimentale studies and is not detailed in models of immune response. Consequently, the influence of macrophage – virus interactions on the infection resolution is unknown. Here, we propose an original model of the immune response centred on the macrophage – virus interactions. We represent all macrophage infectious statuses, their immune functions, and the interactions between innate and adaptive responses taking into account the cytokines regulations. We use the model to study the relative influence of macrophage – virus interactions on the infection resolution by a multivatiate sensitivity analysis. Then, we explore the influence of macrophage immune functions by considering two levels of host susceptibility and viral virulence. We conclude that both repilication rate of the virus and host capacity to synthetize anti-viral cytokines are key for infection resolution.

National audience; Here we aim at providing a general framework capturing the dynamics of co-circulating influenza subtypes and to characterize replacement and co-existence conditions. We propose a time-continuous metapopulation stochastic model for three co-circulating subtypes incorporating three key processes of influenza dynamics: antigenic drift, seasonal forcing in transmission and temporary full cross-immunity. The model realism is enhanced by considering three age-classes with specific contact rates based on data. The network underlying the metapopulation structure comprises major cities in the world coupled through real passengers flows. The impact of key determinants (basic reproductive ratio – R0, antigenic drift rate, and seasonality) was explored through computer-intensive simulations based on parameters consistent with recent estimations. For realistic mean antigenic drift rates, resident subtypes can be replaced even by a new virus with a smaller R0. When R0=1.5 for all subtypes and assuming a strong seasonality, the probability of replacement tends to 1 for a rapidly evolving new subtype whatever the value of the antigenic drift rate for the resident subtypes. Co-existence of the three subtypes can be obtained for plausible ranges of parameter values. Besides the interpretation of the replacement or coexistence in terms of R0 and antigenic drift rate, our approach highlights the need of additional hypotheses on mechanisms avoiding the extinction of a new strain introduced in a system with two strains at equilibrium.

National audience

National audience

FILTREX : Un logiciel convivial pour la microbiologie alimentaire prévisionnelle.Modélisation dynamique de la croissance ou décroissance de populations bactériennes. Journées de microbiologie de l'INRA

International audience; At the 6th ICPMF Gauchi et al. (2009) considered the issue of the identification of complex microbiological dynamic systems and the possibility offered by particle nonlinear filtering to tackle this problem. As the computations involved in this identification approach are rather sophisticated, it is crucial for microbiologists to have access to a user-friendly software for managing them. We present in this 7th ICPMF the FILTREX software, based on Matlab language (Bidot et al., 2009) for reaching several objectives in the predictive microbiology context.

National audience

aeres : C-ACTI; International audience; no abstract

La tremblante, une encéphalopathie spongiforme transmissible ou « maladie à prions », est d’un intérêt majeur dans les troupeaux ovins à cause de son fort impact économique et des campagnes réglementaires d’éradication auxquelles elle est soumise. Ses mécanismes de transmission ne sont encore que partiellement connus. Sa caractéristique principale est un facteur de susceptibilité génétique, lié au polymorphisme du gène PrP codant pour la protéine prion. La combinaison d’une longue période d’incubation infectieuse et l’absence de diagnostic ante-mortem font des cas cliniques les seules observations possibles de la maladie, très rapidement suivies de la mort de l’animal ou de son abattage. La modélisation est donc un outil de choix pour étudier la propagation de la tremblante dans un troupeau ovin. Pour sa flexibilité et son caractère intuitif, nous avons opté pour un modèle individu-centré stochastique, un modèle informatique relevant de l’approche multi-agents. Comme nous disposions de bonnes données de suivi pour un troupeau in- fecté (Langlade, INRA Toulouse), nous avons choisi de ne modéliser que les processus épidémiologiques, contamination et incubation. La démographie est directement issue des données. Nous avons ainsi développé un simulateur sous Java doté d’une interface graphique conviviale. Les données de suivi de troupeau nous ont permis d’estimer les paramètres épidémiologiques du modèle individu-centré, grâce à une méthode d’optimisation reposant sur un algorithme de recherche aléatoire par écart-type adaptatif. Les premiers résultats sont encourageants. Les valeurs des paramètres obtenues sont réalistes et en accord avec de précédentes études. Elles soulignent une forte diminution de la susceptibilité des animaux avant l’âge de un an. Les durées d’incubation mises en évidence sont peu variables, entre individus et entre génotypes. Il nous semble intéressant de prolonger cette étude, en affinant dans un premier temps nos résultats, puis en explorant plus avant différentes hypothèses et scénarios biologiques. Une analyse de sensibilité serait pertinente dans ce contexte. Une autre perspective serait de développer le simulateur en introduisant les processus démographiques dans le modèle, ce qui le rendrait moins dépendant aux données. Nous obtiendrions ainsi un outil de simulation convivial et flexible

International audience; Background: In each country, authorities tried to control the BSE epidemic mainly by feed ban laws. Some questions remain such as the existence of infection through other ways than the artificial food,the efficiency of the main feed ban laws, the initial and the current intensity of the infection and the extinction time of the epidemic.

T cell is one of the most important cells in specific immunity. In order to devise a tool for understanding and predicting some mechanisms of the immune system, a model for T cell response is proposed. The T lymphocyte activation by the recognition of a peptide carried by an antigen presenting cell is an essential step of this immune response. T cell activation was modelled by a system of ordinary differential equations of chemical kinetics type, representing the temporal evolution of the concentrations of different proteins of the lymphocyte (TCR/CD3, CD28, CD69, CD25, IL-2). In order to model a variable quantity of antigen in the organism, a model for virus proliferation was also written, similar to prey predator models. A system of ordinary differential equations thus obtains, involving a specific virus, its target cells (uninfected or infected), and cytotoxic T cell action. A coupling of these two models (T cell activation and virus proliferation) allows simulating the T cell specific response to a viral infection..; Le lymphocyte T est une cellule clé de l'immunité spécifique. Dans le but de créer un outil de compréhension et de prédiction de certains mécanismes immunitaires, une modélisation de la réponse immunitaire du lymphocyte T est proposée. L'activation du lymphocyte par la reconnaissance d'un peptide apprêté par une cellule présentatrice d'antigène est une étape essentielle de cette réponse immunitaire. Cette activation a été modélisée par un système d'équations différentielles ordinaires, de type cinétique chimique, représentant l'évolution temporelle des concentrations de différentes protéines lymphocytaires (TCR/CD3, CD28, CD69, CD25, IL-2). Afin de considérer une quantité d'antigène variable dans l'organisme, un modèle de prolifération virale a aussi été écrit, basé sur des exemples de modèles proies/prédateurs, obtenant ainsi un système d'équations différentielles ordinaires mettant en jeu un virus donné, les cellules cibles du virus, saines ou infectées, et l'action des lymphocytes T cytotoxiques. Un couplage de ces deux modèles (activation lymphocytaire T et prolifération virale) permet une approche de simulation de la réponse lymphocytaire T spécifique à une infection virale.

T cell is one of the most important cells in specific immunity. In order to devise a tool for understanding and predicting some mechanisms of the immune system, a model for T cell response is proposed. The T lymphocyte activation by the recognition of a peptide carried by an antigen presenting cell is an essential step of this immune response. T cell activation was modelled by a system of ordinary differential equations of chemical kinetics type, representing the temporal evolution of the concentrations of different proteins of the lymphocyte (TCR/CD3, CD28, CD69, CD25, IL-2). In order to model a variable quantity of antigen in the organism, a model for virus proliferation was also written, similar to prey predator models. A system of ordinary differential equations thus obtains, involving a specific virus, its target cells (uninfected or infected), and cytotoxic T cell action. A coupling of these two models (T cell activation and virus proliferation) allows simulating the T cell specific response to a viral infection.., Le lymphocyte T est une cellule clé de l'immunité spécifique. Dans le but de créer un outil de compréhension et de prédiction de certains mécanismes immunitaires, une modélisation de la réponse immunitaire du lymphocyte T est proposée. L'activation du lymphocyte par la reconnaissance d'un peptide apprêté par une cellule présentatrice d'antigène est une étape essentielle de cette réponse immunitaire. Cette activation a été modélisée par un système d'équations différentielles ordinaires, de type cinétique chimique, représentant l'évolution temporelle des concentrations de différentes protéines lymphocytaires (TCR/CD3, CD28, CD69, CD25, IL-2). Afin de considérer une quantité d'antigène variable dans l'organisme, un modèle de prolifération virale a aussi été écrit, basé sur des exemples de modèles proies/prédateurs, obtenant ainsi un système d'équations différentielles ordinaires mettant en jeu un virus donné, les cellules cibles du virus, saines ou infectées, et l'action des lymphocytes T cytotoxiques. Un couplage de ces deux modèles (activation lymphocytaire T et prolifération virale) permet une approche de simulation de la réponse lymphocytaire T spécifique à une infection virale.