Machine learning for spatial environmental data. Theory, applications and software

The book presents the state of the art in machine learning algorithms (artificial neural networks of different architectures, support vector machines, etc.) as applied to the classification and mapping of spatially distributed environmental data. Basic geostatistical algorithms are presented as well. New trends in machine learning and their application to spatial data are given, and real case studies based on environmental and pollution data are carried out. The book provides a CD-ROM with the Machine Learning Office software, including sample sets of data, that will allow both students and researchers to put the concepts rapidly to practice.

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.

EuroDia: a beta-cell gene expression resource.

Type 2 diabetes mellitus (T2DM) is a major disease affecting nearly 280 million people worldwide. Whilst the pathophysiological mechanisms leading to disease are poorly understood, dysfunction of the insulin-producing pancreatic beta-cells is key event for disease development. Monitoring the gene expression profiles of pancreatic beta-cells under several genetic or chemical perturbations has shed light on genes and pathways involved in T2DM. The EuroDia database has been established to build a unique collection of gene expression measurements performed on beta-cells of three organisms, namely human, mouse and rat. The Gene Expression Data Analysis Interface (GEDAI) has been developed to support this database. The quality of each dataset is assessed by a series of quality control procedures to detect putative hybridization outliers. The system integrates a web interface to several standard analysis functions from R/Bioconductor to identify differentially expressed genes and pathways. It also allows the combination of multiple experiments performed on different array platforms of the same technology. The design of this system enables each user to rapidly design a custom analysis pipeline and thus produce their own list of genes and pathways. Raw and normalized data can be downloaded for each experiment. The flexible engine of this database (GEDAI) is currently used to handle gene expression data from several laboratory-run projects dealing with different organisms and platforms. Database URL: http://eurodia.vital-it.ch.

Analyse systématique et permanente de la délinquance sérielle : place des statistiques criminelles ; apport des approches situationnelles pour un système de classification ; perspectives en matière de coopération

Résumé de la thèse L'évolution des systèmes policiers donne une place prépondérante à l'information et au renseignement. Cette transformation implique de développer et de maintenir un ensemble de processus permanent d'analyse de la criminalité, en particulier pour traiter des événements répétitifs ou graves. Dans une organisation aux ressources limitées, le temps consacré au recueil des données, à leur codification et intégration, diminue le temps disponible pour l'analyse et la diffusion de renseignements. Les phases de collecte et d'intégration restent néanmoins indispensables, l'analyse n'étant pas possible sur des données volumineuses n'ayant aucune structure. Jusqu'à présent, ces problématiques d'analyse ont été abordées par des approches essentiellement spécialisées (calculs de hot-sports, data mining, ...) ou dirigées par un seul axe (par exemple, les sciences comportementales). Cette recherche s'inscrit sous un angle différent, une démarche interdisciplinaire a été adoptée. L'augmentation continuelle de la quantité de données à analyser tend à diminuer la capacité d'analyse des informations à disposition. Un bon découpage (classification) des problèmes rencontrés permet de délimiter les analyses sur des données pertinentes. Ces classes sont essentielles pour structurer la mémoire du système d'analyse. Les statistiques policières de la criminalité devraient déjà avoir répondu à ces questions de découpage de la délinquance (classification juridique). Cette décomposition a été comparée aux besoins d'un système de suivi permanent dans la criminalité. La recherche confirme que nos efforts pour comprendre la nature et la répartition du crime se butent à un obstacle, à savoir que la définition juridique des formes de criminalité n'est pas adaptée à son analyse, à son étude. Depuis près de vingt ans, les corps de police de Suisse romande utilisent et développent un système de classification basé sur l'expérience policière (découpage par phénomène). Cette recherche propose d'interpréter ce système dans le cadre des approches situationnelles (approche théorique) et de le confronter aux données « statistiques » disponibles pour vérifier sa capacité à distinguer les formes de criminalité. La recherche se limite aux cambriolages d'habitations, un délit répétitif fréquent. La théorie des opportunités soutien qu'il faut réunir dans le temps et dans l'espace au minimum les trois facteurs suivants : un délinquant potentiel, une cible intéressante et l'absence de gardien capable de prévenir ou d'empêcher le passage à l'acte. Ainsi, le délit n'est possible que dans certaines circonstances, c'est-à-dire dans un contexte bien précis. Identifier ces contextes permet catégoriser la criminalité. Chaque cas est unique, mais un groupe de cas montre des similitudes. Par exemple, certaines conditions avec certains environnements attirent certains types de cambrioleurs. Deux hypothèses ont été testées. La première est que les cambriolages d'habitations ne se répartissent pas uniformément dans les classes formées par des « paramètres situationnels » ; la deuxième que des niches apparaissent en recoupant les différents paramètres et qu'elles correspondent à la classification mise en place par la coordination judiciaire vaudoise et le CICOP. La base de données vaudoise des cambriolages enregistrés entre 1997 et 2006 par la police a été utilisée (25'369 cas). Des situations spécifiques ont été mises en évidence, elles correspondent aux classes définies empiriquement. Dans une deuxième phase, le lien entre une situation spécifique et d'activité d'un auteur au sein d'une même situation a été vérifié. Les observations réalisées dans cette recherche indiquent que les auteurs de cambriolages sont actifs dans des niches. Plusieurs auteurs sériels ont commis des délits qui ne sont pas dans leur niche, mais le nombre de ces infractions est faible par rapport au nombre de cas commis dans la niche. Un système de classification qui correspond à des réalités criminelles permet de décomposer les événements et de mettre en place un système d'alerte et de suivi « intelligent ». Une nouvelle série dans un phénomène sera détectée par une augmentation du nombre de cas de ce phénomène, en particulier dans une région et à une période donnée. Cette nouvelle série, mélangée parmi l'ensemble des délits, ne serait pas forcément détectable, en particulier si elle se déplace. Finalement, la coopération entre les structures de renseignement criminel opérationnel en Suisse romande a été améliorée par le développement d'une plateforme d'information commune et le système de classification y a été entièrement intégré.

arrayMap 2014: an updated cancer genome resource.

Somatic copy number aberrations (CNA) represent a mutation type encountered in the majority of cancer genomes. Here, we present the 2014 edition of arrayMap (http://www.arraymap.org), a publicly accessible collection of pre-processed oncogenomic array data sets and CNA profiles, representing a vast range of human malignancies. Since the initial release, we have enhanced this resource both in content and especially with regard to data mining support. The 2014 release of arrayMap contains more than 64,000 genomic array data sets, representing about 250 tumor diagnoses. Data sets included in arrayMap have been assembled from public repositories as well as additional resources, and integrated by applying custom processing pipelines. Online tools have been upgraded for a more flexible array data visualization, including options for processing user provided, non-public data sets. Data integration has been improved by mapping to multiple editions of the human reference genome, with the majority of the data now being available for the UCSC hg18 as well as GRCh37 versions. The large amount of tumor CNA data in arrayMap can be freely downloaded by users to promote data mining projects, and to explore special events such as chromothripsis-like genome patterns.

Orthogonal organic and aqueous-based washes of tissue sections to enhance protein sensitivity by MALDI imaging mass spectrometry.

Imaging mass spectrometry (IMS) is an emergent and innovative approach for measuring the composition, abundance and regioselectivity of molecules within an investigated area of fixed dimension. Although providing unprecedented molecular information compared with conventional MS techniques, enhancement of protein signature by IMS is still necessary and challenging. This paper demonstrates the combination of conventional organic washes with an optimized aqueous-based buffer for tissue section preparation before matrix-assisted laser desorption/ionization (MALDI) IMS of proteins. Based on a 500 mM ammonium formate in water-acetonitrile (9:1; v/v, 0.1% trifluororacetic acid, 0.1% Triton) solution, this buffer wash has shown to significantly enhance protein signature by profiling and IMS (~fourfold) when used after organic washes (70% EtOH followed by 90% EtOH), improving the quality and number of ion images obtained from mouse kidney and a 14-day mouse fetus whole-body tissue sections, while maintaining a similar reproducibility with conventional tissue rinsing. Even if some protein losses were observed, the data mining has demonstrated that it was primarily low abundant signals and that the number of new peaks found is greater with the described procedure. The proposed buffer has thus demonstrated to be of high efficiency for tissue section preparation providing novel and complementary information for direct on-tissue MALDI analysis compared with solely conventional organic rinsing.

Reverse immunology approach for the identification of CD8 T-cell-defined antigens: advantages and hurdles.

One of the challenges of tumour immunology remains the identification of strongly immunogenic tumour antigens for vaccination. Reverse immunology, that is, the procedure to predict and identify immunogenic peptides from the sequence of a gene product of interest, has been postulated to be a particularly efficient, high-throughput approach for tumour antigen discovery. Over one decade after this concept was born, we discuss the reverse immunology approach in terms of costs and efficacy: data mining with bioinformatic algorithms, molecular methods to identify tumour-specific transcripts, prediction and determination of proteasomal cleavage sites, peptide-binding prediction to HLA molecules and experimental validation, assessment of the in vitro and in vivo immunogenic potential of selected peptide antigens, isolation of specific cytolytic T lymphocyte clones and final validation in functional assays of tumour cell recognition. We conclude that the overall low sensitivity and yield of every prediction step often requires a compensatory up-scaling of the initial number of candidate sequences to be screened, rendering reverse immunology an unexpectedly complex approach.

The future key role of LC-high-resolution-MS analyses in clinical laboratories: a focus on quantification.

For the last decade, high-resolution (HR)-MS has been associated with qualitative analyses while triple quadrupole MS has been associated with routine quantitative analyses. However, a shift of this paradigm is taking place: quantitative and qualitative analyses will be increasingly performed by HR-MS, and it will become the common 'language' for most mass spectrometrists. Most analyses will be performed by full-scan acquisitions recording 'all' ions entering the HR-MS with subsequent construction of narrow-width extracted-ion chromatograms. Ions will be available for absolute quantification, profiling and data mining. In parallel to quantification, metabotyping will be the next step in clinical LC-MS analyses because it should help in personalized medicine. This article is aimed to help analytical chemists who perform targeted quantitative acquisitions with triple quadrupole MS make the transition to quantitative and qualitative analyses using HR-MS. Guidelines for the acceptance criteria of mass accuracy and for the determination of mass extraction windows in quantitative analyses are proposed.

Current approaches and challenges for the metabolite profiling of complex natural extracts.

Metabolite profiling is critical in many aspects of the life sciences, particularly natural product research. Obtaining precise information on the chemical composition of complex natural extracts (metabolomes) that are primarily obtained from plants or microorganisms is a challenging task that requires sophisticated, advanced analytical methods. In this respect, significant advances in hyphenated chromatographic techniques (LC-MS, GC-MS and LC-NMR in particular), as well as data mining and processing methods, have occurred over the last decade. Together, these tools, in combination with bioassay profiling methods, serve an important role in metabolomics for the purposes of both peak annotation and dereplication in natural product research. In this review, a survey of the techniques that are used for generic and comprehensive profiling of secondary metabolites in natural extracts is provided. The various approaches (chromatographic methods: LC-MS, GC-MS, and LC-NMR and direct spectroscopic methods: NMR and DIMS) are discussed with respect to their resolution and sensitivity for extract profiling. In addition the structural information that can be generated through these techniques or in combination, is compared in relation to the identification of metabolites in complex mixtures. Analytical strategies with applications to natural extracts and novel methods that have strong potential, regardless of how often they are used, are discussed with respect to their potential applications and future trends.

AWSum-Combining Classification with Knowledge Aquisition

Many classifiers achieve high levels of accuracy but have limited applicability in real world situations because they do not lead to a greater understanding or insight into the^way features influence the classification. In areas such as health informatics a classifier that clearly identifies the influences on classification can be used to direct research and formulate interventions. This research investigates the practical applications of Automated Weighted Sum, (AWSum), a classifier that provides accuracy comparable to other techniques whilst providing insight into the data. This is achieved by calculating a weight for each feature value that represents its influence on the class value. The merits of this approach in classification and insight are evaluated on a Cystic Fibrosis and Diabetes datasets with positive results.

Rpe65 microarrays, from genes to phosphorylated proteins

Purpose:To describe a novel in silico method to gather and analyze data from high-throughput heterogeneous experimental procedures, i.e. gene and protein expression arrays. Methods:Each microarray is assigned to a database which handles common data (names, symbols, antibody codes, probe IDs, etc.). Links between informations are automatically generated from knowledge obtained in freely accessible databases (NCBI, Swissprot, etc). Requests can be made from any point of entry and the displayed result is fully customizable. Results:The initial database has been loaded with two sets of data: a first set of data originating from an Affymetrix-based retinal profiling performed in an RPE65 knock-out mouse model of Leber's congenital amaurosis. A second set of data generated from a Kinexus microarray experiment done on the retinas from the same mouse model has been added. Queries display wild type versus knock out expressions at several time points for both genes and proteins. Conclusions:This freely accessible database allows for easy consultation of data and facilitates data mining by integrating experimental data and biological pathways.

Multi-well fungal co-culture for de novo metabolite-induction in time-series studies based on untargeted metabolomics.

The induction of fungal metabolites by fungal co-cultures grown on solid media was explored using multi-well co-cultures in 2 cm diameter Petri dishes. Fungi were grown in 12-well plates to easily and rapidly obtain the large number of replicates necessary for employing metabolomic approaches. Fungal culture using such a format accelerated the production of metabolites by several weeks compared with using the large-format 9 cm Petri dishes. This strategy was applied to a co-culture of a Fusarium and an Aspergillus strain. The metabolite composition of the cultures was assessed using ultra-high pressure liquid chromatography coupled to electrospray ionisation and time-of-flight mass spectrometry, followed by automated data mining. The de novo production of metabolites was dramatically increased by nutriment reduction. A time-series study of the induction of the fungal metabolites of interest over nine days revealed that they exhibited various induction patterns. The concentrations of most of the de novo induced metabolites increased over time. However, interesting patterns were observed, such as with the presence of some compounds only at certain time points. This result indicates the complexity and dynamic nature of fungal metabolism. The large-scale production of the compounds of interest was verified by co-culture in 15 cm Petri dishes; most of the induced metabolites of interest (16/18) were found to be produced as effectively as on a small scale, although not in the same time frames. Large-scale production is a practical solution for the future production, identification and biological evaluation of these metabolites.

Bioinformatics-driven identification and examination of candidate genes for non-alcoholic fatty liver disease.

ObjectiveCandidate genes for non-alcoholic fatty liver disease (NAFLD) identified by a bioinformatics approach were examined for variant associations to quantitative traits of NAFLD-related phenotypes.Research Design and MethodsBy integrating public database text mining, trans-organism protein-protein interaction transferal, and information on liver protein expression a protein-protein interaction network was constructed and from this a smaller isolated interactome was identified. Five genes from this interactome were selected for genetic analysis. Twenty-one tag single-nucleotide polymorphisms (SNPs) which captured all common variation in these genes were genotyped in 10,196 Danes, and analyzed for association with NAFLD-related quantitative traits, type 2 diabetes (T2D), central obesity, and WHO-defined metabolic syndrome (MetS).Results273 genes were included in the protein-protein interaction analysis and EHHADH, ECHS1, HADHA, HADHB, and ACADL were selected for further examination. A total of 10 nominal statistical significant associations (P<0.05) to quantitative metabolic traits were identified. Also, the case-control study showed associations between variation in the five genes and T2D, central obesity, and MetS, respectively. Bonferroni adjustments for multiple testing negated all associations.ConclusionsUsing a bioinformatics approach we identified five candidate genes for NAFLD. However, we failed to provide evidence of associations with major effects between SNPs in these five genes and NAFLD-related quantitative traits, T2D, central obesity, and MetS.

Automated scoring of AFLPs using RawGeno v 2.0, a free R CRAN library.

Amplified Fragment Length Polymorphisms (AFLPs) are a cheap and efficient protocol for generating large sets of genetic markers. This technique has become increasingly used during the last decade in various fields of biology, including population genomics, phylogeography, and genome mapping. Here, we present RawGeno, an R library dedicated to the automated scoring of AFLPs (i.e., the coding of electropherogram signals into ready-to-use datasets). Our program includes a complete suite of tools for binning, editing, visualizing, and exporting results obtained from AFLP experiments. RawGeno can either be used with command lines and program analysis routines or through a user-friendly graphical user interface. We describe the whole RawGeno pipeline along with recommendations for (a) setting the analysis of electropherograms in combination with PeakScanner, a program freely distributed by Applied Biosystems; (b) performing quality checks; (c) defining bins and proceeding to scoring; (d) filtering nonoptimal bins; and (e) exporting results in different formats.

The connectome viewer toolkit: an open source framework to manage, analyze, and visualize connectomes.

Advanced neuroinformatics tools are required for methods of connectome mapping, analysis, and visualization. The inherent multi-modality of connectome datasets poses new challenges for data organization, integration, and sharing. We have designed and implemented the Connectome Viewer Toolkit - a set of free and extensible open source neuroimaging tools written in Python. The key components of the toolkit are as follows: (1) The Connectome File Format is an XML-based container format to standardize multi-modal data integration and structured metadata annotation. (2) The Connectome File Format Library enables management and sharing of connectome files. (3) The Connectome Viewer is an integrated research and development environment for visualization and analysis of multi-modal connectome data. The Connectome Viewer's plugin architecture supports extensions with network analysis packages and an interactive scripting shell, to enable easy development and community contributions. Integration with tools from the scientific Python community allows the leveraging of numerous existing libraries for powerful connectome data mining, exploration, and comparison. We demonstrate the applicability of the Connectome Viewer Toolkit using Diffusion MRI datasets processed by the Connectome Mapper. The Connectome Viewer Toolkit is available from http://www.cmtk.org/