Machine learning for geospatial data : algorithms, software tools and case studies

Résumé Cette thèse est consacrée à l'analyse, la modélisation et la visualisation de données environnementales à référence spatiale à l'aide d'algorithmes d'apprentissage automatique (Machine Learning). L'apprentissage automatique peut être considéré au sens large comme une sous-catégorie de l'intelligence artificielle qui concerne particulièrement le développement de techniques et d'algorithmes permettant à une machine d'apprendre à partir de données. Dans cette thèse, les algorithmes d'apprentissage automatique sont adaptés pour être appliqués à des données environnementales et à la prédiction spatiale. Pourquoi l'apprentissage automatique ? Parce que la majorité des algorithmes d'apprentissage automatiques sont universels, adaptatifs, non-linéaires, robustes et efficaces pour la modélisation. Ils peuvent résoudre des problèmes de classification, de régression et de modélisation de densité de probabilités dans des espaces à haute dimension, composés de variables informatives spatialisées (« géo-features ») en plus des coordonnées géographiques. De plus, ils sont idéaux pour être implémentés en tant qu'outils d'aide à la décision pour des questions environnementales allant de la reconnaissance de pattern à la modélisation et la prédiction en passant par la cartographie automatique. Leur efficacité est comparable au modèles géostatistiques dans l'espace des coordonnées géographiques, mais ils sont indispensables pour des données à hautes dimensions incluant des géo-features. Les algorithmes d'apprentissage automatique les plus importants et les plus populaires sont présentés théoriquement et implémentés sous forme de logiciels pour les sciences environnementales. Les principaux algorithmes décrits sont le Perceptron multicouches (MultiLayer Perceptron, MLP) - l'algorithme le plus connu dans l'intelligence artificielle, le réseau de neurones de régression généralisée (General Regression Neural Networks, GRNN), le réseau de neurones probabiliste (Probabilistic Neural Networks, PNN), les cartes auto-organisées (SelfOrganized Maps, SOM), les modèles à mixture Gaussiennes (Gaussian Mixture Models, GMM), les réseaux à fonctions de base radiales (Radial Basis Functions Networks, RBF) et les réseaux à mixture de densité (Mixture Density Networks, MDN). Cette gamme d'algorithmes permet de couvrir des tâches variées telle que la classification, la régression ou l'estimation de densité de probabilité. L'analyse exploratoire des données (Exploratory Data Analysis, EDA) est le premier pas de toute analyse de données. Dans cette thèse les concepts d'analyse exploratoire de données spatiales (Exploratory Spatial Data Analysis, ESDA) sont traités selon l'approche traditionnelle de la géostatistique avec la variographie expérimentale et selon les principes de l'apprentissage automatique. La variographie expérimentale, qui étudie les relations entre pairs de points, est un outil de base pour l'analyse géostatistique de corrélations spatiales anisotropiques qui permet de détecter la présence de patterns spatiaux descriptible par une statistique. L'approche de l'apprentissage automatique pour l'ESDA est présentée à travers l'application de la méthode des k plus proches voisins qui est très simple et possède d'excellentes qualités d'interprétation et de visualisation. Une part importante de la thèse traite de sujets d'actualité comme la cartographie automatique de données spatiales. Le réseau de neurones de régression généralisée est proposé pour résoudre cette tâche efficacement. Les performances du GRNN sont démontrées par des données de Comparaison d'Interpolation Spatiale (SIC) de 2004 pour lesquelles le GRNN bat significativement toutes les autres méthodes, particulièrement lors de situations d'urgence. La thèse est composée de quatre chapitres : théorie, applications, outils logiciels et des exemples guidés. Une partie importante du travail consiste en une collection de logiciels : Machine Learning Office. Cette collection de logiciels a été développée durant les 15 dernières années et a été utilisée pour l'enseignement de nombreux cours, dont des workshops internationaux en Chine, France, Italie, Irlande et Suisse ainsi que dans des projets de recherche fondamentaux et appliqués. Les cas d'études considérés couvrent un vaste spectre de problèmes géoenvironnementaux réels à basse et haute dimensionnalité, tels que la pollution de l'air, du sol et de l'eau par des produits radioactifs et des métaux lourds, la classification de types de sols et d'unités hydrogéologiques, la cartographie des incertitudes pour l'aide à la décision et l'estimation de risques naturels (glissements de terrain, avalanches). Des outils complémentaires pour l'analyse exploratoire des données et la visualisation ont également été développés en prenant soin de créer une interface conviviale et facile à l'utilisation. Machine Learning for geospatial data: algorithms, software tools and case studies Abstract The thesis is devoted to the analysis, modeling and visualisation of spatial environmental data using machine learning algorithms. In a broad sense machine learning can be considered as a subfield of artificial intelligence. It mainly concerns with the development of techniques and algorithms that allow computers to learn from data. In this thesis machine learning algorithms are adapted to learn from spatial environmental data and to make spatial predictions. Why machine learning? In few words most of machine learning algorithms are universal, adaptive, nonlinear, robust and efficient modeling tools. They can find solutions for the classification, regression, and probability density modeling problems in high-dimensional geo-feature spaces, composed of geographical space and additional relevant spatially referenced features. They are well-suited to be implemented as predictive engines in decision support systems, for the purposes of environmental data mining including pattern recognition, modeling and predictions as well as automatic data mapping. They have competitive efficiency to the geostatistical models in low dimensional geographical spaces but are indispensable in high-dimensional geo-feature spaces. The most important and popular machine learning algorithms and models interesting for geo- and environmental sciences are presented in details: from theoretical description of the concepts to the software implementation. The main algorithms and models considered are the following: multi-layer perceptron (a workhorse of machine learning), general regression neural networks, probabilistic neural networks, self-organising (Kohonen) maps, Gaussian mixture models, radial basis functions networks, mixture density networks. This set of models covers machine learning tasks such as classification, regression, and density estimation. Exploratory data analysis (EDA) is initial and very important part of data analysis. In this thesis the concepts of exploratory spatial data analysis (ESDA) is considered using both traditional geostatistical approach such as_experimental variography and machine learning. Experimental variography is a basic tool for geostatistical analysis of anisotropic spatial correlations which helps to understand the presence of spatial patterns, at least described by two-point statistics. A machine learning approach for ESDA is presented by applying the k-nearest neighbors (k-NN) method which is simple and has very good interpretation and visualization properties. Important part of the thesis deals with a hot topic of nowadays, namely, an automatic mapping of geospatial data. General regression neural networks (GRNN) is proposed as efficient model to solve this task. Performance of the GRNN model is demonstrated on Spatial Interpolation Comparison (SIC) 2004 data where GRNN model significantly outperformed all other approaches, especially in case of emergency conditions. The thesis consists of four chapters and has the following structure: theory, applications, software tools, and how-to-do-it examples. An important part of the work is a collection of software tools - Machine Learning Office. Machine Learning Office tools were developed during last 15 years and was used both for many teaching courses, including international workshops in China, France, Italy, Ireland, Switzerland and for realizing fundamental and applied research projects. Case studies considered cover wide spectrum of the real-life low and high-dimensional geo- and environmental problems, such as air, soil and water pollution by radionuclides and heavy metals, soil types and hydro-geological units classification, decision-oriented mapping with uncertainties, natural hazards (landslides, avalanches) assessments and susceptibility mapping. Complementary tools useful for the exploratory data analysis and visualisation were developed as well. The software is user friendly and easy to use.

Genetic factors of obesity and eating disorders: Copy number variations (CNV) involving SH2B1

Context : It is now clearly shown that genetic factors in association with environment play a key role in obesity and eating disorders. This project studies the clinical symptoms and molecular abnormalities in patients carrying a strong hereditary predisposition to obesity and eating behavior disorders. We have previously published the association between the 16:29.5-30.1 deletion and a very penetrant form of morbid obesity and macrocephaly. We have also demonstrated the association between the reciprocal 16:29.5-30.1 duplication and underweight and small head circumference. These 2 studies demonstrate that gene dosage of one or several genes in this region regulates BMI as well as brain growth. At present, there are no data pointing towards particular candidate genes. We are currently investigating a second non-overlapping recurrent CNV encompassing SH2B1, upstream of the aforementioned rearrangement. SNPs in this gene have been associated with BMI in GWAS studies and mice models confirmed this association. Bokuchova et al have reported an association between deletions encompassing this gene and severe early onset obesity, as well as insulin resistance. We are currently collecting and analyzing data to fully characterize the phenotype and the transcriptional patterns associated with this rearrangement. Aims : 1. Identify carriers of any CNVs in the greater 16p11.2 region (between 16:28MB and 32MB) in the EGG consortium. 2. Perform association studies between SNPs in the greater 16p11.2 region (16:28-32MB) and anthropometric measures with adjusted "locus-wide significance", to identify or prioritize candidate genes potentially driving the association observed in patients with the CNVs (and thus worthy of further validation and sequencing). 3. Explore associations between GSV genome-wide and brain volume. 4. Explore relationship between brain volumes (whole brain and regional for those who underwent brain MRI), head circumference and BMI. 5. Extrapolate this procedure to other regions covered by the Metabochip. Methods : - Examine and collect clinical informations, as well as molecular informations in these patients. - Analysis of MRI data in children and adults with BMI > 2SD. Compare changes to MRI data obtained in patients with monogenic forms of obesity (data from Lausanne study) and to underweight (BMI<-2SD) individuals from EGG. - Test whether opposite extremes of the phenotypic distribution may be highly informative Expected results : This is a highly focused study, pertaining to approximately 1 0/00 of the human genome. Yet it is clear that if successful, the lessons learned from this study could be extrapolated to other segments of the genome and would need validation and replication by additional studies. Altogether they will contribute to further explore the missing heritability and point to etiologic genes and pathways underlying these important health burdens.

Mutations in DNAH5 account for only 15% of a non-preselected cohort of patients with primary ciliary dyskinesia.

BACKGROUND: Primary ciliary dyskinesia (PCD) is characterised by recurrent infections of the upper respiratory airways (nose, bronchi, and frontal sinuses) and randomisation of left-right body asymmetry. To date, PCD is mainly described with autosomal recessive inheritance and mutations have been found in five genes: the dynein arm protein subunits DNAI1, DNAH5 and DNAH11, the kinase TXNDC3, and the X-linked retinitis pigmentosa GTPase regulator RPGR. METHODS: We screened 89 unrelated individuals with PCD for mutations in the coding and splice site regions of the gene DNAH5 by denaturing high performance liquid chromatography (DHPLC) and sequencing. Patients were mainly of European origin and were recruited without any phenotypic preselection. RESULTS: We identified 18 novel (nonsense, splicing, small deletion and missense) and six previously described mutations. Interestingly, these DNAH5 mutations were mainly associated with outer + inner dyneins arm ultrastructural defects (50%). CONCLUSION: Overall, mutations on both alleles of DNAH5 were identified in 15% of our clinically heterogeneous cohort of patients. Although genetic alterations remain to be identified in most patients, DNAH5 is to date the main PCD gene.

Genetic variants influencing circulating lipid levels and risk of coronary artery disease.

OBJECTIVE: Genetic studies might provide new insights into the biological mechanisms underlying lipid metabolism and risk of CAD. We therefore conducted a genome-wide association study to identify novel genetic determinants of low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and triglycerides. METHODS AND RESULTS: We combined genome-wide association data from 8 studies, comprising up to 17 723 participants with information on circulating lipid concentrations. We did independent replication studies in up to 37 774 participants from 8 populations and also in a population of Indian Asian descent. We also assessed the association between single-nucleotide polymorphisms (SNPs) at lipid loci and risk of CAD in up to 9 633 cases and 38 684 controls. We identified 4 novel genetic loci that showed reproducible associations with lipids (probability values, 1.6×10(-8) to 3.1×10(-10)). These include a potentially functional SNP in the SLC39A8 gene for HDL-C, an SNP near the MYLIP/GMPR and PPP1R3B genes for LDL-C, and at the AFF1 gene for triglycerides. SNPs showing strong statistical association with 1 or more lipid traits at the CELSR2, APOB, APOE-C1-C4-C2 cluster, LPL, ZNF259-APOA5-A4-C3-A1 cluster and TRIB1 loci were also associated with CAD risk (probability values, 1.1×10(-3) to 1.2×10(-9)). CONCLUSIONS: We have identified 4 novel loci associated with circulating lipids. We also show that in addition to those that are largely associated with LDL-C, genetic loci mainly associated with circulating triglycerides and HDL-C are also associated with risk of CAD. These findings potentially provide new insights into the biological mechanisms underlying lipid metabolism and CAD risk.

Estimating parameters for generalized mass action models using constraint propagation.

As modern molecular biology moves towards the analysis of biological systems as opposed to their individual components, the need for appropriate mathematical and computational techniques for understanding the dynamics and structure of such systems is becoming more pressing. For example, the modeling of biochemical systems using ordinary differential equations (ODEs) based on high-throughput, time-dense profiles is becoming more common-place, which is necessitating the development of improved techniques to estimate model parameters from such data. Due to the high dimensionality of this estimation problem, straight-forward optimization strategies rarely produce correct parameter values, and hence current methods tend to utilize genetic/evolutionary algorithms to perform non-linear parameter fitting. Here, we describe a completely deterministic approach, which is based on interval analysis. This allows us to examine entire sets of parameters, and thus to exhaust the global search within a finite number of steps. In particular, we show how our method may be applied to a generic class of ODEs used for modeling biochemical systems called Generalized Mass Action Models (GMAs). In addition, we show that for GMAs our method is amenable to the technique in interval arithmetic called constraint propagation, which allows great improvement of its efficiency. To illustrate the applicability of our method we apply it to some networks of biochemical reactions appearing in the literature, showing in particular that, in addition to estimating system parameters in the absence of noise, our method may also be used to recover the topology of these networks.

Non-viral ocular gene therapy: potential ocular therapeutic avenues.

Non-viral vectors for potential gene replacement and therapy have been developed in order to overcome the drawbacks of viral vectors. The diversity of non-viral vectors allows for a wide range of various products, flexibility of application, ease of use, low-cost of production and enhanced "genomic" safety. Using non-viral strategies, oligonucleotides (ODNs) can be delivered naked (less efficient) or entrapped in cationic lipids, polymers or peptides forming slow release delivery systems, which can be adapted according to the organ targeted and the therapy purposes. Tissue and cell internalization can be further enhanced by changing by physical or chemical means. Moreover, a specific vector can be selected according to disease course and intensity of manifestations fulfilling specific requirements such as the duration of drug release and its level along with cells and tissues specific targeting. From accumulating knowledge and experience, it appears that combination of several non-viral techniques may increase the efficacy and ensure the safety of these evolving and interesting gene therapy strategies.

Environmental data mining and modeling based on machine learning algorithms and geostatistics

The paper presents some contemporary approaches to spatial environmental data analysis. The main topics are concentrated on the decision-oriented problems of environmental spatial data mining and modeling: valorization and representativity of data with the help of exploratory data analysis, spatial predictions, probabilistic and risk mapping, development and application of conditional stochastic simulation models. The innovative part of the paper presents integrated/hybrid model-machine learning (ML) residuals sequential simulations-MLRSS. The models are based on multilayer perceptron and support vector regression ML algorithms used for modeling long-range spatial trends and sequential simulations of the residuals. NIL algorithms deliver non-linear solution for the spatial non-stationary problems, which are difficult for geostatistical approach. Geostatistical tools (variography) are used to characterize performance of ML algorithms, by analyzing quality and quantity of the spatially structured information extracted from data with ML algorithms. Sequential simulations provide efficient assessment of uncertainty and spatial variability. Case study from the Chernobyl fallouts illustrates the performance of the proposed model. It is shown that probability mapping, provided by the combination of ML data driven and geostatistical model based approaches, can be efficiently used in decision-making process. (C) 2003 Elsevier Ltd. All rights reserved.

C4 Photosynthesis evolved in grasses via parallel adaptive genetic changes.

Phenotypic convergence is a widespread and well-recognized evolutionary phenomenon. However, the responsible molecular mechanisms remain often unknown mainly because the genes involved are not identified. A well-known example of physiological convergence is the C4 photosynthetic pathway, which evolved independently more than 45 times [1]. Here, we address the question of the molecular bases of the C4 convergent phenotypes in grasses (Poaceae) by reconstructing the evolutionary history of genes encoding a C4 key enzyme, the phosphoenolpyruvate carboxylase (PEPC). PEPC genes belong to a multigene family encoding distinct isoforms of which only one is involved in C4 photosynthesis [2]. By using phylogenetic analyses, we showed that grass C4 PEPCs appeared at least eight times independently from the same non-C4 PEPC. Twenty-one amino acids evolved under positive selection and converged to similar or identical amino acids in most of the grass C4 PEPC lineages. This is the first record of such a high level of molecular convergent evolution, illustrating the repeatability of evolution. These amino acids were responsible for a strong phylogenetic bias grouping all C4 PEPCs together. The C4-specific amino acids detected must be essential for C4 PEPC enzymatic characteristics, and their identification opens new avenues for the engineering of the C4 pathway in crops.

Within- and among-population impact of genetic erosion on adult fitness-related traits in the European tree frog Hyla arborea.

Assessing in wild populations how fitness is impacted by inbreeding and genetic drift is a major goal for conservation biology. An approach to measure the detrimental effects of inbreeding on fitness is to estimate correlations between molecular variation and phenotypic performances within and among populations. Our study investigated the effect of individual multilocus heterozygosity on body size, body condition and reproductive investment of males (that is, chorus attendance) and females (that is, clutch mass and egg size) in both small fragmented and large non-fragmented populations of European tree frog (Hyla arborea). Because adult size and/or condition and reproductive investment are usually related, genetic erosion may have detrimental effects directly on reproductive investment, and also on individual body size and condition that in turn may affect reproductive investment. We confirmed that the reproductive investment was highly size-dependent for both sexes. Larger females invested more in offspring production, and larger males attended the chorus in the pond more often. Our results did not provide evidence for a decline in body size, condition and reproductive effort with decreased multilocus heterozygosity both within and among populations. We showed that the lack of heterozygosity-fitness correlations within populations probably resulted from low inbreeding levels (inferior to ca. 20% full-sib mating rate), even in the small fragmented populations. The detrimental effects of fixation load were either low in adults or hidden by environmental variation among populations. These findings will be useful to design specific management actions to improve population persistence.

Genetic aberrations in pediatric follicular lymphoma

Background: Pediatric follicular lymphoma (FL) is a rare disease that differs from its adult counterpart both genetically and clinically. Excluding pediatric FL with IRF4-translocation, the genetic events associated with pediatric FL have not yet been defined. Objectives: The aim of this study was to perform a complete genetic characterization of IRF4-translocation negative pediatric follicular lymphomas to elucidate the genetic profile of these rare pediatric cases and determine common genetic alterations that could be associated to this phenotype. Design/Methods: We applied array-comparative genomic hybridization and molecular inversion probe assay adapted to formalin-fixed paraffin-embedded tissues from 18 patients aged £18 years diagnosed with FL. With the exception of one case with only focal involvement by lymphoma, the tumor cell content exceeded 50% in the evaluable samples. Eleven of 18 patients were treated according to NHL-BFM group multicenter trials whereas the remaining according to different protocols. All lacked t(14;18) translocation. Mutational analysis of TNFRSF14 gene was performed in 17 cases. Results: Only six pediatric cases displayed chromosomal imbalances, with gain/amplification of 6pter-p24.3 (including IRF4) and deletion/ copy number neutral-loss of heterozygosity in 1p36 (including TNFRSF14) being the most frequent alterations. Sequencing of the candidate gene TNFRSF14 at 1p36.32 showed nine mutations in seven cases. Conclusion: Combination of molecular and genetic features differentiated a recurrent pattern of genomic imbalances as well as of TNFRSF14 mutations in pediatric FL which together with other genetic alterations distinguishes two subsets of pediatric follicular lymphomas. The first group shows genomic aberrations and is associated with more aggressive histopathologic and clinical features. The second group lacks genetic alterations detectable with the present approaches and is associated with a more limited disease. Despite the absence of genomic aberrations, these cases resembled FL by their histopathological features.

Nestmate recognition and levels of aggression are not altered by changes in genetic diversity in a socially polymorphic ant

The ability to distinguish nestmates from foreign individuals is central to the functioning of insect societies. In ants, workers from multiple-queen colonies are often less aggressive than workers from single-queen ones. In line with this observation, it has been hypothesized that workers from multiple-queen colonies have less precise recognition abilities than workers from single-queen ones because their colonies contain genetically more diverse individuals, which results in a broader template of recognition cues. Here, we assessed the impact of social structure ( queen number) variation on nestmate recognition and aggression in a large population of the socially polymorphic ant Formica selysi. We staged unilateral aggression tests on the nest surface. Workers from single-and multiple-queen colonies had good nestmate recognition ability and did not differ significantly in their level of aggression towards foreign, immobilized workers ( cue-bearers). In particular, workers from multiple-queen colonies efficiently recognized non-nestmates despite the higher genetic diversity in their colony. Cue-bearers from single- and multiple-queen colonies elicited similar reactions. However, the level of aggression was higher between than within social forms, suggesting that workers detect a signal that is specific to the colony social structure. Finally, the level of aggression was not correlated with the genetic distance between colonies. Overall, we found no evidence for the hypothesis that the presence of multiple breeders in the same colony decreases recognition abilities and found no simple relationship between genetic diversity and aggression level. (c) 2007 The Association for the Study of Animal Behaviou

Division of labor in insect societies : genetic components and physiological regulation

The shift from solitary to social organisms constitutes one of the major transitions in evolution. The highest level of sociality is found in social insects (ants, termites and some species of bees and wasps). Division of labor is central to the organization of insect societies and is thought to be at the root of their ecological success. There are two main levels of division of labor in social insect colonies. The first relates to reproduction and involves the coexistence of queen and worker castes: while reproduction is usually monopolized by one or several queens, functionally sterile workers perform all the tasks to maintain the colony, such as nest building, foraging or brood care. The second level of division of labor, relating to such non-reproductive duties, is characterized by the performance of different tasks or roles by different groups of workers. This PhD aims to better understand the mechanisms underlying division of labor in insect societies, by investigating how genes and physiology influence caste determination and worker behavior in ants. In the first axis of this PhD, we studied the nature of genetic effects on division of labor. We used the Argentine ant Linepithema humile to conduct controlled crosses in the laboratory, which revealed the existence of non-additive genetic effects, such as parent-of-origin and genetic compatibility effects, on caste determination and worker behavior. In the second axis, we focused on the physiological regulation of division of labor. Using Pogonomyrmex seed- harvester ants, we performed experimental manipulation of hibernation, hormonal treatments, gene expression analyses and protein quantification to identify the physiological pathways regulating maternal effects on caste determination. Finally, comparing gene expression between nurses and foragers allowed us to reveal the association between vitellogenin and worker behavior in Pogonomyrmex ants. This PhD provides important insights into the role of genes and physiology in the regulation of division of labor in social insect colonies, helping to better understand the organization, evolution and ecological success of insect societies. - L'une des principales transitions évolutives est le passage de la vie solitaire à la vie sociale. La socialité atteint son paroxysme chez les insectes sociaux que sont les fourmis, les termites et certaines espèces d'abeilles et de guêpes. La division du travail est la clé de voûte de l'organisation de ces sociétés d'insectes et la raison principale de leur succès écologique. La division du travail s'effectue à deux niveaux dans les colonies d'insectes sociaux. Le premier niveau concerne la reproduction et implique la coexistence de deux castes : les reines et les ouvrières. Tandis que la reproduction est le plus souvent monopolisée par une ou plusieurs reines, les ouvrières stériles effectuent les tâches nécessaires au bon fonctionnement de la colonie, telles que la construction du nid, la recherche de nourriture ou le soin au couvain. Le second niveau de division du travail, qui concerne les tâches autres que la reproduction, implique la réalisation de différents travaux par différents groupes d'ouvrières. Le but de ce doctorat est de mieux comprendre les mécanismes sous-jacents de la division du travail dans les sociétés d'insectes en étudiant comment les gènes et la physiologie influencent la détermination de la caste et le comportement des ouvrières chez les fourmis. Dans le premier axe de ce doctorat, nous avons étudié la nature des influences génétiques sur la division du travail. Nous avons utilisé la fourmi d'Argentine, Linepithema humile, pour effectuer des croisements contrôlés en laboratoire. Cette méthode nous a permis de révéler l'existence d'influences génétiques non additives, telles que des influences dépendantes de l'origine parentale ou des effets de compatibilité génétique, sur la détermination de la caste et le comportement des ouvrières. Dans le second axe, nous nous sommes intéressés à la régulation physiologique de la division du travail. Nous avons utilisé des fourmis moissonneuses du genre Pogonomyrmex pour effectuer des hibernations artificieHes, des traitements hormonaux, des analyses d'expression de gènes et des mesures de vitellogénine, ce qui nous a permis d'identifier les mécanismes physiologiques régulant les effets maternels sur la détermination de la caste. Enfin, la comparaison d'expression de gènes entre nourrices et fourrageuses suggère un rôle de la vitellogénine dans la régulation du comportement des ouvrières chez les fourmis moissonneuses. En détaillant les influences des gènes et de la physiologie dans la régulation de la division du travail dans les colonies d'insectes sociaux, ce doctorat fournit d'importantes informations permettant de mieux comprendre l'organisation, l'évolution et le succès écologique des sociétés d'insectes.

Official Positions for FRAX® clinical regarding prior fractures from Joint Official Positions Development Conference of the International Society for Clinical Densitometry and International Osteoporosis Foundation on FRAX®.

The 2010 Position Development Conference addressed four questions related to the impact of previous fractures on 10-year fracture risk as calculated by FRAX(®). To address these questions, PubMed was searched on the keywords "fracture, epidemiology, osteoporosis." Titles of retrieved articles were reviewed for an indication that risk for future fracture was discussed. Abstracts of these articles were reviewed for an indication that one or more of the questions listed above was discussed. For those that did, the articles were reviewed in greater detail to extract the findings and to find additional past work and citing works that also bore on the questions. The official positions and the supporting literature review are presented here. FRAX(®) underestimates fracture probability in persons with a history of multiple fractures (good, A, W). FRAX(®) may underestimate fracture probability in individuals with prevalent severe vertebral fractures (good, A, W). While there is evidence that hip, vertebral, and humeral fractures appear to confer greater risk of subsequent fracture than fractures at other sites, quantification of this incremental risk in FRAX(®) is not possible (fair, B, W). FRAX(®) may underestimate fracture probability in individuals with a parental history of non-hip fragility fracture (fair, B, W). Limitations of the methodology include performance by a single reviewer, preliminary review of the literature being confined to titles, and secondary review being limited to abstracts. Limitations of the evidence base include publication bias, overrepresentation of persons of European descent in the published studies, and technical differences in the methods used to identify prevalent and incident fractures. Emerging topics for future research include fracture epidemiology in non-European populations and men, the impact of fractures in family members other than parents, and the genetic contribution to fracture risk.

Population genetics of Capercaillie (Tetrao urogallus) in the Jura and the Pyrenees: a non-invasive approach to avian conservation genetics

RÉSUMÉ Le Grand tétras est un galliforme de montagne apparenté au faisan et au tétras lyre. Il est distribué de manière continue à travers la toundra et les montagnes de moyenne altitude en Europe de l'ouest. Toutefois, les populations d'Europe de l'ouest ont subi un déclin constant au cours des derniers siècles. Les causes de ce déclin sont probablement liées à l'activité humaine, telle .que l'élevage ou le tourisme, qui ont engendré une modification et une fragmentation de l'habitat de l'espèce. Malheureusement, les populations soumises à de forts déclins démographiques peuvent subir des effets génétiques (augmentation de la consanguinité et perte de diversité génétique) pouvant diminuer leur potentiel de reproduction et conduire irrémédiablement à l'extinction. Cette thèse présente les analyses conduites dans le but d'estimer l'impact du déclin démographique des populations de Grand tétras sur l'étendue et la distribution de leur variabilité génétique dans le Jura et dans les Pyrénées. Du fait de la législation locale protégeant les tétraonidés en général, mais également en raison de la biologie très cryptique du Grand tétras, l'ensemble des analyses de cette étude a été réalisé à partir de matériel génétique extrait des fientes (ou échantillonnage génétique non invasif). Dans la première partie de l'étude, je détaille les protocoles d'extraction. d'ADN et d'amplification par PCR modifiés à partir des protocoles classiques utilisant des échantillons conventionnels, riches en ADN. L'utilisation d'ADN fécal impose des contraintes dues à la mauvaise qualité et à la faible quantité du matériel génétique à disposition dans les fientes. Ces contraintes ont pu être partiellement contournées en réalisant des répétitions multiples du génotypage afin d'obtenir un degré de fiabilité suffisante. J'ai également analysé les causes de la dégradation de l'ADN dans les excréments. Parmi les causes les plus communes, telles que l'activité bactérienne, l'hydrolyse spontanée et la dégradation enzymatique par les DNases libres, c'est ce dernier facteur qui apparaît comme étant la cause majeure et la plus rapide responsable de la dégradation de la qualité des échantillons. La rapidité de l'action enzymatique suggère que les plans d'échantillonnages de excréments sur le terrain pourraient être optimisés en les réalisant dans des conditions climatiques froides et sèches, favorisant ainsi l'inhibition des DNases. La seconde partie de la thèse est une étude par simulation visant à déterminer la capacité du logiciel Structure à identifier les structures génétiques complexes et hiérarchiques fréquemment rencontrées dans les populations naturelles, et ce en utilisant différents types de marqueurs génétiques. Les troisième et quatrième parties de cette thèse décrivent le statut génétique des populations résiduelles du Jura et des Pyrénées à partir de l'analyse de 11 loci microsatellites. Nous n'avons pas pu mettre en évidence dans les deux populations des effets liés à la consanguinité ou à la réduction de la diversité génétique. De plus, la différenciation génétique entre les patches d'habitats favorables reste modérée et corrélée à la distance géographique, ce qui suggère que la dispersion d'individus entre les patches a été importante au moins pendant ces dernières générations. La comparaison des paramètres de la diversité génétique avec ceux d'autres populations de Grand tétras, ou d'autres espèces proches, indique que la population du Jura a retenu une proportion importante de sa diversité originelle. Ces résultats suggèrent que le déclin récent des populations a jusqu'ici eu un impact modéré sur les facteurs génétiques et que ces populations semblent avoir conservé le potentiel génétique nécessaire à leur survie à long terme. Finalement, en cinquième partie, l'analyse de l'apparentement entre les mâles qui participent à la parade sur les places de chant (leks) indique que ces derniers sont distribués en agrégats de manière non aléatoire, préférentiellement entre individus apparentés. De plus, la corrélation entre les distances génétique et géographique entre les leks est en accord avec les motifs d'isolement par la distance mis en évidence à d'autres niveaux hiérarchiques (entre patches d'habitat et populations), ainsi qu'avec les études menées sur d'autres espèces ayant choisi ce même système de reproduction. En conclusion, cette première étude basée uniquement sur de l'ADN nucléaire aviaire extrait à partir de fèces a fourni des informations nouvelles qui n'auraient pas pu être obtenues par une méthode d'observation sur le terrain ou d'échantillonnage génétique classique. Aucun oiseau n'a été dérangé ou capturé, et les résultats sont comparables à d'autres études concernant des espèces proches. Néanmoins, la taille de ces populations approche des niveaux au-dessous desquels la survie à long terme est fortement incertaine. La persistance de la diversité génétique pour les prochaines générations reste en conséquence liée à la survie des adultes et à une reprise du succès de la reproduction. ABSTRACT Capercaillie (Tetrao urogallus) is a large grouse that is continuously distributed across the tundra and the mid-high mountains of Western Europe. However, the populations in Western Europe have been showing a constant decline during the last decades. The causes for this decline are possibly related to human activities, such as cattle breeding and tourism that have both led to habitat modification and fragmentation. Unfortunately, populations that have undergone drastic demographic bottlenecks often go through genetic processes of inbreeding and loss of diversity that decrease their fitness and eventually lead to extinction. This thesis presents the investigations conducted to estimate the impact of the demographic decline of capercaillie populations on the extent and distribution of their genetic variability in the Jura and in the Pyrenees mountains. Because grouse are protected by wildlife legislation, and also because of the cryptic behaviour of capercaillie, all DNA material used in this study was extracted from faeces (non-invasive genetic sampling). In the first part of my thesis, I detail the protocols of DNA extraction and PCR amplification adapted from classical methods using conventional DNA-rich samples. The use of faecal DNA imposes specific constraints due to the low quantity and the highly degraded genetic material available. These constraints are partially overcome by performing multiple genotyping repetitions to obtain sufficient reliability. I also investigate the causes of DNA degradation in faeces. Among the main degraders, namely bacterial activity, spontaneous hydrolysis, and free-¬DNase activities, the latter was pointed out as the most important according to our experiments. These enzymes degrade DNA very rapidly, and, as a consequence, faeces sampling schemes must be planned preferably in cold and dry weather conditions, allowing for enzyme activity inhibition. The second part of the thesis is a simulation study aiming to assess the capacity of the software Structure to detect population structure in hierarchical models relevant to situations encountered in wild populations, using several genetic markers. The methods implemented in Structure appear efficient in detecting the highest hierarchical structure. The third and fourth parts of the thesis describe the population genetics status of the remaining Jura and Pyrenees populations using 11 microsatellite loci. In either of these populations, no inbreeding nor reduced genetic diversity was detected. Furthermore, the genetic differentiation between patches defined by habitat suitability remains moderate and correlated with geographical distance, suggesting that significant dispersion between patches was at work at least until the last generations. The comparison of diversity indicators with other species or other populations of capercaillie indicate that population in the Jura has retained a large part of its original genetic diversity. These results suggest that the recent decline has had so forth a moderate impact on• genetic factors and that these populations might have retained the potential for long term survival, if the decline is stopped. Finally, in the fifth part, the analysis of relatedness between males participating in the reproduction parade, or lek, indicate that capercaillie males, like has been shown for some other grouse species, gather on leks• among individuals that are more related than the average of the population. This pattern appears to be due to both population structure and kin-association. As a conclusion, this first study relying exclusively on nuclear DNA extracted from faeces has provided novel information that was not available through field observation or classical genetic sampling. No bird has been captured or disturbed, and the results are consistent with other studies of closely related species. However, the size of these populations is approaching thresholds below which long-term survival is unlikely. The persistence of genetic diversity for the forthcoming generations remains therefore bond to adult survival and to the increase of reproduction success.

Organisation of genetic variation in multinucleate arbuscular mycorrhizal fungi

Les Champignons Endomycorhiziens Arbusculaires (CEA) forment une symbiose racinaire avec environ 80% des espèces connues de plantes vasculaires. Ils occupent une position écologique très importante liée aux bénéfices qu'ils confèrent aux plantes. Des études moléculaires effectuées sur des gènes ribosomaux ont révélé un très grand polymorphisme, tant à l'intérieur des espèces qu'entre celles-ci. Ces champignons étant coenocytiques et multinucléés, l'organisation de cette variabilité génétique intraspécifique pourrait avoir différentes origines. Ce travail se propose d'examiner l'organisation et l'évolution de cette variabilité. Sur la base de fossiles, l'existence des CEA remonte à au moins 450 millions d'années. Cette symbiose peut donc être considérée comme ancienne. Les premières données moléculaires n'indiquant pas de reproduction sexuée, une hypothèse fut élaborée stipulant que les CEA seraient des asexués ancestraux. La première partie de cette thèse (chapitre 2) met en évidence l'existence de recombinaison dans différents CEA mais montre également que celle-ci est insuffisante pour purger les mutations accumulées. La reproduction étant essentiellement asexuée, on peut prédire que les nombreux noyaux ont probablement divergé génétiquement. En collaboration avec M. Hijri nous avons pu vérifier cette hypothèse (chapitre 2). Dans le chapitre 3 j'ai cherché à comprendre si le polymorphisme était également présent dans une population naturelle du CEA Glomus intraradices au niveau intraspécifique, ce qui n'avait encore jamais été examiné. En comparant les empreintes génétiques d'individus obtenus chacun à partir d'une spore mise en culture, j'ai clairement démontré que d'importantes différences génétiques existent entre ceux-ci. Un résultat similaire, portant sur des traits quantitatifs d'individus de la même population, a été trouvé par A. Koch. Les deux études en ensemble montre que le polymorphisme génétique dans cette population est suffisamment grand pour être important au niveau écologique. Dans le chapitre 4, j'ai cherché a examiner le polymorphisme des séquences du gène BiP au sein d'un individu. C'est la première étude qui examine la diversité génétique du génome de CEA avec un autre marqueur que l'ADN ribosomique. J'ai trouvé 31 types de séquences différentes du gène BiP issu d'un isolat de G. intraradices mis en culture à partir d'une seule spore. Cette variation n'était pas restreinte à des zones sélectivement neutres du BiP. Mes résultats montrent qu'il y a un grand nombre de variants non-fonctionnels, proportionnellement au faible nombre de copies attendues par noyau. Ceci va dans le sens d'une partition de l'information génétique entre les noyaux.<br/><br/>Arbuscular mycorrhizal fungi (AMF) are root symbionts with about 80% of all known species of vascular land plants. AMF are ecologically important because of the benefits that they confer to plants. Molecular studies on AMF showed that rDNA sequences were highly variable between species and within species. Because AMF are coenocytic and multinucleate there are several possibilities how this intraspecific genetic variation could be organized. Therefore, the organization and evolution of this variation in AMF were investigated in the present work. Based on fossil records the AMF symbiosis has existed for 450 Million years and is therefore considered ancient. First molecular data indicated no evident sexual reproduction and gave rise to the hypothesis that AMF might be ancient asexuals. The first part of this thesis (Chapter 2) shows evidence for recombination in different AMF but also indicates that it has not been frequent enough to purge accumulated mutations. Given asexual reproduction, it has been predicted that the many nuclei in AMF should diverge leading to genetically different nuclei. This hypothesis has been confirmed by an experiment of M. Hijri and is also included in chapter 2 as the results were published together. In chapter 3 I then investigated whether intraspecific genetic variation also exists in a field population of the AMF Glomus intraradices. Comparing genetic fingerprints of individuals derived from single spores I could clearly show that large genetic differences exist. A similar result, based on quantitative genetic traits, was found for the same population by A. Koch. The two studies taken together show that the genetic variation observed in the population is high enough to be of ecological relevance. Lastly, in chapter 4, I investigated within individual genetic variation among BiP gene sequences. It is the first study that has analyzed genetic diversity in the AMF genome in a region of DNA other than rDNA. I found 31 sequence variants of the BiP gene in one G. intraradices isolate that originated from one spore. Genetic variation was not only restricted to selectively neutral parts of BiP. A high number of predicted non-functional variants compared to a likely low number of copies per nucleus indicated that functional genetic information might even be partitioned among nuclei. The results of this work contribute to our understanding of potential evolutionary strategies of ancient asexuals, they also suggest that genetic differences in a population might be ecologically relevant and they show that this variation even occurs in functional regions of the AMF genome.