Bayesian data assimilation

This thesis addresses data assimilation, which typically refers to the estimation of the state of a physical system given a model and observations, and its application to short-term precipitation forecasting. A general introduction to data assimilation is given, both from a deterministic and' stochastic point of view. Data assimilation algorithms are reviewed, in the static case (when no dynamics are involved), then in the dynamic case. A double experiment on two non-linear models, the Lorenz 63 and the Lorenz 96 models, is run and the comparative performance of the methods is discussed in terms of quality of the assimilation, robustness "in the non-linear regime and computational time. Following the general review and analysis, data assimilation is discussed in the particular context of very short-term rainfall forecasting (nowcasting) using radar images. An extended Bayesian precipitation nowcasting model is introduced. The model is stochastic in nature and relies on the spatial decomposition of the rainfall field into rain "cells". Radar observations are assimilated using a Variational Bayesian method in which the true posterior distribution of the parameters is approximated by a more tractable distribution. The motion of the cells is captured by a 20 Gaussian process. The model is tested on two precipitation events, the first dominated by convective showers, the second by precipitation fronts. Several deterministic and probabilistic validation methods are applied and the model is shown to retain reasonable prediction skill at up to 3 hours lead time. Extensions to the model are discussed.

The pharmacology of adrenomedullin 2/intermedin

Adrenomedullin 2 (AM2) or intermedin is a member of the calcitonin gene-related peptide (CGRP)/calcitonin family of peptides and was discovered in 2004. Unlike other members of this family, no unique receptor has yet been identified for it. It is extensively distributed throughout the body. It causes hypotension when given peripherally, but when given into the CNS, it increases blood pressure and causes sympathetic activation. It also increases prolactin release, is anti-diuretic and natriuretic and reduces food intake. Whilst its effects resemble those of AM, it is frequently more potent. Some characterization of AM2 has been done on molecularly defined receptors; the existing data suggest that it preferentially activates the AM receptor formed from calcitonin receptor-like receptor and receptor activity modifying protein 3. On this complex, its potency is generally equivalent to that of AM. There is no known receptor-activity where it is more potent than AM. In tissues and in animals it is frequently antagonised by CGRP and AM antagonists; however, situations exist in which an AM2 response is maintained even in the presence of supramaximal concentrations of these antagonists. Thus, there is a partial mismatch between the pharmacology seen in tissues and that on cloned receptors. The only AM2 antagonists are peptide fragments, and these have limited selectivity. It remains unclear as to whether novel AM2 receptors exist or whether the mismatch in pharmacology can be explained by factors such as metabolism. © 2011 The British Pharmacological Society.

Gaussian process quantile regression using expectation propagation

Direct quantile regression involves estimating a given quantile of a response variable as a function of input variables. We present a new framework for direct quantile regression where a Gaussian process model is learned, minimising the expected tilted loss function. The integration required in learning is not analytically tractable so to speed up the learning we employ the Expectation Propagation algorithm. We describe how this work relates to other quantile regression methods and apply the method on both synthetic and real data sets. The method is shown to be competitive with state of the art methods whilst allowing for the leverage of the full Gaussian process probabilistic framework.

Projected sequential Gaussian processes:a C++ tool for interpolation of large datasets with heterogeneous noise

Heterogeneous datasets arise naturally in most applications due to the use of a variety of sensors and measuring platforms. Such datasets can be heterogeneous in terms of the error characteristics and sensor models. Treating such data is most naturally accomplished using a Bayesian or model-based geostatistical approach; however, such methods generally scale rather badly with the size of dataset, and require computationally expensive Monte Carlo based inference. Recently within the machine learning and spatial statistics communities many papers have explored the potential of reduced rank representations of the covariance matrix, often referred to as projected or fixed rank approaches. In such methods the covariance function of the posterior process is represented by a reduced rank approximation which is chosen such that there is minimal information loss. In this paper a sequential Bayesian framework for inference in such projected processes is presented. The observations are considered one at a time which avoids the need for high dimensional integrals typically required in a Bayesian approach. A C++ library, gptk, which is part of the INTAMAP web service, is introduced which implements projected, sequential estimation and adds several novel features. In particular the library includes the ability to use a generic observation operator, or sensor model, to permit data fusion. It is also possible to cope with a range of observation error characteristics, including non-Gaussian observation errors. Inference for the covariance parameters is explored, including the impact of the projected process approximation on likelihood profiles. We illustrate the projected sequential method in application to synthetic and real datasets. Limitations and extensions are discussed. © 2010 Elsevier Ltd.

Investigation of molecular mechanisms of LKB1 function in Dictyostelium development and stress responses and the function of superoxide dismutase C (SodC) in chemotaxis

The serine/threonine kinase LKB1 is a regulator of critical events including development and stress responses in metazoans. The current study was undertaken to determine the function of LKB1 in Dictyostelium . During multicellular development and in response to stress insult, an apparent increase in the DdLKB1 kinase activity was observed. Depletion of DdLKB1 with a knockdown construct led to aberrant development; a severe reduction in prespore cell differentiation and a precocious induction of prestalk cells, which were reminiscent of cells lacking GSK3, a well known cell-fate switch. Furthermore, DdLKB1 depleted cells displayed lower GSK3 activity than wild type cells in response to cAMP stimulation during development and failed to activate AMPK, a well known LKB1 target in mammals, in response to cAMP and stress insults. These results suggest that DdLKB1 positively regulates both GSK3 and AMPK during Dictyostelium development, and DdLKB1 is necessary for AMPK activation during stress response regulation. No apparent GSK3 activation was observed in response to stress insults. Spatial and temporal regulation of phosphatidylinositol-(3,4,5)-triphosphate (PIP3) along the membrane of polarized cells is important for efficient chemotaxis. A REMI screen for PIP3 suppressors in the absence of stimulation led to the identification of SodC as PIP3 regulator. Consistent with their higher PIP3 levels, sodC− cells showed defects in chemotaxis and exhibited higher intra-cellular superoxide levels. Protein localization studies along with observations from GPI specific PI-PLC treatment of wild-type cells suggested that SodC is a GPI anchored outer-membrane protein. SodC showed superoxide dismutase activity in vitro, and motility defects of sodC− cells can be rescued by expressing the intact SodC but not by the mutant SodC, which has point mutations that affect its dismutase function. Treatment of sodC− cells with LY294002, a pharmacological inhibitor of PI3K, partially rescued the polarization and chemoattractant sensing defects but not motility defects. Consistent with increased intracellular superoxide levels, sodC − cells also exhibited higher basal Ras activity, an upstream regulator of PI3K, which can be suppressed by a cell permeable superoxide scavenger, XTT, indicating that SodC is important in regulation of intracellular superoxide levels thereby regulating the Ras activity and PIP3 levels at the membrane.

Investigation of molecular mechanisms of lkb1 function in dictyostelium development and stress responses and the function of superoxide dismutase c (sodc) in chemotaxis

The serine/threonine kinase LKB1 is a regulator of critical events including development and stress responses in metazoans. The current study was undertaken to determine the function of LKB1 in Dictyostelium. During multicellular development and in response to stress insult, an apparent increase in the DdLKB1 kinase activity was observed. Depletion of DdLKB1 with a knockdown construct led to aberrant development; a severe reduction in prespore cell differentiation and a precocious induction of prestalk cells, which were reminiscent of cells lacking GSK3, a well known cell-fate switch. Furthermore, DdLKB1 depleted cells displayed lower GSK3 activity than wild type cells in response to cAMP stimulation during development and failed to activate AMPK, a well known LKB1 target in mammals, in response to cAMP and stress insults. These results suggest that DdLKB1 positively regulates both GSK3 and AMPK during Dictyostelium development, and DdLKB1 is necessary for AMPK activation during stress response regulation. No apparent GSK3 activation was observed in response to stress insults. Spatial and temporal regulation of phosphatidylinositol-(3,4,5)-triphosphate (PIP3) along the membrane of polarized cells is important for efficient chemotaxis. A REMI screen for PIP3 suppressors in the absence of stimulation led to the identification of SodC as PIP3 regulator. Consistent with their higher PIP3 levels, sodC- cells showed defects in chemotaxis and exhibited higher intra-cellular superoxide levels. Protein localization studies along with observations from GPI specific PI-PLC treatment of wild-type cells suggested that SodC is a GPI anchored outer-membrane protein. SodC showed superoxide dismutase activity in vitro, and motility defects of sodC- cells can be rescued by expressing the intact SodC but not by the mutant SodC, which has point mutations that affect its dismutase function. Treatment of sodC- cells with LY294002, a pharmacological inhibitor of PI3K, partially rescued the polarization and chemoattractant sensing defects but not motility defects. Consistent with increased intracellular superoxide levels, sodC- cells also exhibited higher basal Ras activity, an upstream regulator of PI3K, which can be suppressed by a cell permeable superoxide scavenger, XTT, indicating that SodC is important in regulation of intracellular superoxide levels thereby regulating the Ras activity and PIP3 levels at the membrane.

Cosmologie, mythologie et récit historique dans la tradition orale des Algonquins de Kitcisakik

Thèse numérisée par la Direction des bibliothèques de l'Université de Montréal.

Cosmologie, mythologie et récit historique dans la tradition orale des Algonquins de Kitcisakik

Thèse numérisée par la Direction des bibliothèques de l'Université de Montréal.

In Vitro modelling of RSV infection and cytopathogenesis in well-differentiated human primary airway epithelial cells (WD-PAECs).

The choice of model used to study human respiratory syncytial virus (RSV) infection is extremely important. RSV is a human pathogen that is exquisitely adapted to infection of human hosts. Rodent models, such as mice and cotton rats, are semi-permissive to RSV infection and do not faithfully reproduce hallmarks of RSV disease in humans. Furthermore, immortalized airway-derived cell lines, such as HEp-2, BEAS-2B, and A549 cells, are poorly representative of the complexity of the respiratory epithelium. The development of a well-differentiated primary pediatric airway epithelial cell models (WD-PAECs) allows us to simulate several hallmarks of RSV infection of infant airways. They therefore represent important additions to RSV pathogenesis modeling in human-relevant tissues. The following protocols describe how to culture and differentiate both bronchial and nasal primary pediatric airway epithelial cells and how to use these cultures to study RSV cytopathogenesis.

EGFR interacts with the fusion protein of respiratory syncytial virus strain 2-20 and mediates infection and mucin expression.

Respiratory syncytial virus (RSV) is the major cause of viral lower respiratory tract illness in children. In contrast to the RSV prototypic strain A2, clinical isolate RSV 2-20 induces airway mucin expression in mice, a clinically relevant phenotype dependent on the fusion (F) protein of the RSV strain. Epidermal growth factor receptor (EGFR) plays a role in airway mucin expression in other systems; therefore we hypothesized that the RSV 2-20 F protein stimulates EGFR signaling. Infection of cells with chimeric strains RSV A2-2-20F and A2-2-20GF or over-expression of 2-20 F protein resulted in greater phosphorylation of EGFR than infection with RSV A2 or over-expression of A2 F, respectively. Chemical inhibition of EGFR signaling or knockdown of EGFR resulted in diminished infectivity of RSV A2-2-20F but not RSV A2. Over-expression of EGFR enhanced the fusion activity of 2-20 F protein in trans. EGFR co-immunoprecipitated most efficiently with RSV F proteins derived from “mucogenic” strains. RSV 2-20 F and EGFR co-localized in H292 cells, and A2-2-20GF-induced MUC5AC expression was ablated by EGFR inhibitors in these cells. Treatment of BALB/c mice with the EGFR inhibitor erlotinib significantly reduced the amount of RSV A2-2-20F-induced airway mucin expression. Our results demonstrate that RSV F interacts with EGFR in a strain-specific manner, EGFR is a co-factor for infection, and EGFR plays a role in RSV-induced mucin expression, suggesting EGFR is a potential target for RSV disease.

Journées Rémi des 23 et 24 novembre 1998 - Réseau de contrôle microbiologique (Rémi) et surveillance microbiologique

Ce document est un rapport de synthèse des présentations faites lors des journées REMI, tenues à Nantes les 23 et 24 novembre 1998. Les exposés de la première journée sont principalement destinés aux actions visant à améliorer le fonctionnement du réseau REMI. Ils concernent les sujets suivants : bilan de la mise en oeuvre du réseau de contrôle REMI en 1998 (nouvelle version), bilan de l'état d'avancement du plan qualité surveillance microbiologique et du programme Malthus (méthodologie, normalisation), approches préliminaires pour la détermination des fréquences d'échantillonnage en vue de la surveillance régulière des zones de production classées. La seconde journée est dédiée aux études conduites par les laboratoires côtiers, dont les procédures sont définies par la charte de surveillance microbiologique, ou réalisées par le Laboratoire de Microbiologie. Le dispositif de programmation des études de surveillance et de saisie des données dans la base Quadrige est présenté. Les sujets abordés se rapportent aux thèmes suivants : impact de la plaisance, des populations aviaires, des rejets agricoles et urbains sur la contamination bactériologique des coquillages ou des eaux littorales, les bactériophages, évaluation de l'importance de la pêche récréative par survol aérien des gisements coquilliers, objectifs et réflexions sur les perspectives d'un système d'évaluation de la qualité des eaux littorales (SEQ Littoral). Le bilan des journées REMI s'analyse à travers les commentaires recueillis lors du tour de table, et le relevé des propositions d'actions.

Etude sanitaire de la zone 17.48 de Fouras. Charente-Maritime

Suite à la demande d’exploitation de la zone de Fouras, n°17.48 par le Comité Régional de la Conchyliculture du Poitou-Charentes (CRCPC), la Direction Départementale des Territoires et de la Mer (DDTM) a demandé la réalisation d’une étude sanitaire en vue du classement de la zone de production de coquillages pour les bivalves fouisseurs (groupe 2). Cette étude réalisée par le Laboratoire Environnement Ressources des Pertuis Charentais (LER/PC) bénéficie d’un financement de la Direction Générale de l 'Alimentation (DGAL). Basée sur les paramètres microbiologiques (Escherichia coli) et chimiques (plomb, mercure, cadmium), l’étude sanitaire a pour objectifs d'estimer la qualité microbiologique et chimique de la zone en vue du classement sanitaire de la zone par l'administration conformément aux exigences du Règlement CE n° 854/2004 ; et de déterminer la stratégie d'échantillonnage à mettre en oeuvre dans le cadre de la surveillance sanitaire régulière de cette zone suite à son classement. L’étude des informations disponibles a permis l’identification de sources de contamination potentielle et la définition d’une stratégie d’échantillonnage. Deux points de suivis ont été positionnés dans les secteurs jugés sensibles aux sources de contamination et ont été échantillonnés du 29 juillet 2011 au 4 octobre 2012. Les concentrations maximales en cadmium, mercure et plomb sont inférieures aux critères chimiques réglementaires et sont compatibles avec un classement A, B ou C. Le suivi microbiologique réalisé indique une qualité B pour chacun des deux points de suivi. La qualité de la zone semble donc homogène, et est estimée B dans son ensemble. Le point « Les Ecussons » est le plus sensible aux contaminations microbiologiques, la surveillance régulière REMI de la zone du Fier d’Ars pour le groupe 2 sera donc basée sur ce point.

Etude sanitaire de la zone 17.52 - Ronce les Bains - Barat, Charente-Maritime

Suite à la demande d'exploitation de la zone de Ronce les Bains – Barat par le Comité Régional de la Conchyliculture du Poitou-Charentes (CRCPC), la Direction Départementale du Territoire et de la Mer (DDTM) a demandé la réalisation d'une étude sanitaire en vue du classement de cette zone de production pour les bivalves fouisseurs (groupe 2). Cette étude réalisée par le Laboratoire Environnement Ressources des Pertuis Charentais bénéficie d'un financement de la Direction Générale de l'Alimentation (DGAL). Basée sur le paramètre microbiologique (Escherichia coli) et chimiques (plomb, mercure, cadmium). L'étude sanitaire a pour objectifs : d'estimer la qualité microbiologique et chimique de la zone en vue du classement sanitaire de la zone par l'administration conformément aux exigences du Règlement CE n° 854/2004 ; et de déterminer la stratégie d'échantillonnage à mettre en oeuvre dans le cadre de la surveillance sanitaire régulière de cette zone suite à son classement. L’étude des informations disponibles a permis l’identification de sources de contamination potentielle et la définition d’une stratégie d’échantillonnage. Trois points de suivi ont été positionnés dans des secteurs jugés sensibles aux sources de contamination, et ont été échantillonnés de février 2011 à janvier 2013. Les concentrations maximales en cadmium, mercure et plomb sont inférieures aux critères chimiques réglementaires et sont compatibles avec un classement A, B ou C. Le suivi microbiologique réalisé indique une qualité B pour chacun des deux points de suivi selon les seuils microbiologiques définis par le règlement (CE) n° 854/2004 (un troisième point a été arrêté au cours de l’étude pour manque de ressource). La qualité de la zone est donc estimée B dans son ensemble. Le point « Ronce » est le plus sensible aux contaminations microbiologiques, la surveillance régulière REMI de la zone de Ronce les Bains – Barat pour le groupe 2 sera donc basée sur ce point.

Etude sanitaire de la zone 17.51 de Bourgeois. Charente-Maritime.

Suite à la demande d’exploitation de la zone de Bourgeois par les professionnels, la Direction Départementale du Territoire et de la Mer (DDTM) a demandé la réalisation d’une étude sanitaire en vue du classement de cette zone de production pour les bivalves fouisseurs (groupe 2 ).Cette étude réalisée par le Laboratoire Environnement ressources des Pertuis Charentais bénéficie d’un financement de la Direction Générale de l’Alimentation (DGAL). Basée sur le paramètre microbiologique (Escherichia coli) et chimiques (plomb, mercure, cadmium), l’étude sanitaire a pour objectifs : d’estimer la qualité microbiologique et chimique de la zone en vue du classement sanitaire de la zone par l’administration conformément aux exigences du Règlement CE n° 854/2004 ; et de déterminer la stratégie d’échantillonnage à mettre en œuvre dans le cadre de la surveillance sanitaire régulière de cette zone suite à son classement. L’étude des informations disponibles a permis l’identification de sources de contamination potentielle et la définition d’une stratégie d’échantillonnage. Trois points de suivi ont été positionnés dans des secteurs jugés sensibles aux sources de contamination et ont été échantillonnés du 21 juillet 2011 au 20 août 2012. Les concentrations maximales en cadmium, mercure et plomb sont inférieures aux critères chimiques réglementaires et sont compatibles avec un classement en A, B ou C. Le suivi microbiologique réalisé indique une qualité B pour chacun des trois points de suivi selon les seuils microbiologiques définis par le règlement (CE) n° 854/2004. La qualité de la zone est donc estimée B dans son ensemble. Le point « Bonnemort » est retenu pour la surveillance régulière REMI du groupe 2 dans la zone n° 17-51 Bourgeois.

Site n°5. The Pertuis Breton (Bay of Biscay) France

1.This report presents the results of a field study conducted in the ECASA test site nOS in the Pertuis Breton, France. The site is located on the Atlantic West coasts. It is open to the bay of Biscay, but is slightly protected against westerly winds. The bay has been exploited by intertidal mussels culture for centuries. 2. Within the bay, mussels (Mytilus edulis) are cultivated either by the traditional pole technique, around the bay or on longlines in the centre of the bay. The area occupied by these longline s represents 250 ha, and the resulting annual production is 1 000 tonnes of mussels. The average depth at mid tide is of 13.8 m. The sediment is sandy, with a small fraction of mud. 3. The site is subject to several regular monitoring through the local implementation of national networks aiming at protecting the environment and marine resources, on pollutants (RNO), microbiological quality of the waters (REMI), phytoplanktonic toxic species (REPHY) and growth and mortality of molluscs (REMORA). Benthic macrofauna was studied in 1976. 4. Five sampling sations were chosen along a line, starting under the longlines, and at distances of 50, 100, 200, and 400 metres from the area cultivated. A reference station was chosen in a different direction at 2300 metres of the cultivated area. Sampling methods are described in the text. _Sediments were sampled for different analyses: grain size, content in organic matter, total organic carbon and nitrogen, and phytic pigments (chlorophyll a and phaeopigments). Redox were measured in cores. The macrofauna living into the sediment was also sampled. The water column was sampled for physical (temperature, transparency) and chemical parametres, including oxygen content, salinity, organic matter, dissolved nitrogen forms, phosphates and silicates. Results from benthic macrofauna surveys indicate that there were no significant differences between the different stations and the reference station, all being classified as slightly disturbed. The bay is submitted to freshwater runoffs from two adjacent rivers. 7. The sediment is slightly modified by the culture of bivalves. Total organic carbon, total nitrogen, Eh values and pheopigments were significantly higher under the trestles than in any other stations. Other stations often did not differ from the reference station. 8. The effects of shellfish culture on the water column were. However, it was observed a small decrease of the food available to the molluscs near the rearing 9. The DEB model was able to describe and predict adequately the growth of oysters, both in the Baie des Veys and in the Loch Creran. The parametres for its use in other environment are given, but a tuning of one parametre should be performed with the help of authors. 10. Among the indicators and models for use in are as of intertidal bivalve culture, it is recommended to use the sediment quality index, TOC (Total Organic Carbon), redox and pheopigments, in surficial sediment, AMBI for the macrofauna, chlorophyll a contents and nitrogen forms in the water column, and models describing the carrying capacity, filtration rate of molluscs, and a DEB model to predict the growth of molluscs.