Ensemble classifiers in remote sensing: a comparative analysis

Dissertation submitted in partial fulfillment of the requirements for the Degree of Master of Science in Geospatial Technologies.

Automatic learning for the classification of chemical reactions and in statistical thermodynamics

This Thesis describes the application of automatic learning methods for a) the classification of organic and metabolic reactions, and b) the mapping of Potential Energy Surfaces(PES). The classification of reactions was approached with two distinct methodologies: a representation of chemical reactions based on NMR data, and a representation of chemical reactions from the reaction equation based on the physico-chemical and topological features of chemical bonds. NMR-based classification of photochemical and enzymatic reactions. Photochemical and metabolic reactions were classified by Kohonen Self-Organizing Maps (Kohonen SOMs) and Random Forests (RFs) taking as input the difference between the 1H NMR spectra of the products and the reactants. The development of such a representation can be applied in automatic analysis of changes in the 1H NMR spectrum of a mixture and their interpretation in terms of the chemical reactions taking place. Examples of possible applications are the monitoring of reaction processes, evaluation of the stability of chemicals, or even the interpretation of metabonomic data. A Kohonen SOM trained with a data set of metabolic reactions catalysed by transferases was able to correctly classify 75% of an independent test set in terms of the EC number subclass. Random Forests improved the correct predictions to 79%. With photochemical reactions classified into 7 groups, an independent test set was classified with 86-93% accuracy. The data set of photochemical reactions was also used to simulate mixtures with two reactions occurring simultaneously. Kohonen SOMs and Feed-Forward Neural Networks (FFNNs) were trained to classify the reactions occurring in a mixture based on the 1H NMR spectra of the products and reactants. Kohonen SOMs allowed the correct assignment of 53-63% of the mixtures (in a test set). Counter-Propagation Neural Networks (CPNNs) gave origin to similar results. The use of supervised learning techniques allowed an improvement in the results. They were improved to 77% of correct assignments when an ensemble of ten FFNNs were used and to 80% when Random Forests were used. This study was performed with NMR data simulated from the molecular structure by the SPINUS program. In the design of one test set, simulated data was combined with experimental data. The results support the proposal of linking databases of chemical reactions to experimental or simulated NMR data for automatic classification of reactions and mixtures of reactions. Genome-scale classification of enzymatic reactions from their reaction equation. The MOLMAP descriptor relies on a Kohonen SOM that defines types of bonds on the basis of their physico-chemical and topological properties. The MOLMAP descriptor of a molecule represents the types of bonds available in that molecule. The MOLMAP descriptor of a reaction is defined as the difference between the MOLMAPs of the products and the reactants, and numerically encodes the pattern of bonds that are broken, changed, and made during a chemical reaction. The automatic perception of chemical similarities between metabolic reactions is required for a variety of applications ranging from the computer validation of classification systems, genome-scale reconstruction (or comparison) of metabolic pathways, to the classification of enzymatic mechanisms. Catalytic functions of proteins are generally described by the EC numbers that are simultaneously employed as identifiers of reactions, enzymes, and enzyme genes, thus linking metabolic and genomic information. Different methods should be available to automatically compare metabolic reactions and for the automatic assignment of EC numbers to reactions still not officially classified. In this study, the genome-scale data set of enzymatic reactions available in the KEGG database was encoded by the MOLMAP descriptors, and was submitted to Kohonen SOMs to compare the resulting map with the official EC number classification, to explore the possibility of predicting EC numbers from the reaction equation, and to assess the internal consistency of the EC classification at the class level. A general agreement with the EC classification was observed, i.e. a relationship between the similarity of MOLMAPs and the similarity of EC numbers. At the same time, MOLMAPs were able to discriminate between EC sub-subclasses. EC numbers could be assigned at the class, subclass, and sub-subclass levels with accuracies up to 92%, 80%, and 70% for independent test sets. The correspondence between chemical similarity of metabolic reactions and their MOLMAP descriptors was applied to the identification of a number of reactions mapped into the same neuron but belonging to different EC classes, which demonstrated the ability of the MOLMAP/SOM approach to verify the internal consistency of classifications in databases of metabolic reactions. RFs were also used to assign the four levels of the EC hierarchy from the reaction equation. EC numbers were correctly assigned in 95%, 90%, 85% and 86% of the cases (for independent test sets) at the class, subclass, sub-subclass and full EC number level,respectively. Experiments for the classification of reactions from the main reactants and products were performed with RFs - EC numbers were assigned at the class, subclass and sub-subclass level with accuracies of 78%, 74% and 63%, respectively. In the course of the experiments with metabolic reactions we suggested that the MOLMAP / SOM concept could be extended to the representation of other levels of metabolic information such as metabolic pathways. Following the MOLMAP idea, the pattern of neurons activated by the reactions of a metabolic pathway is a representation of the reactions involved in that pathway - a descriptor of the metabolic pathway. This reasoning enabled the comparison of different pathways, the automatic classification of pathways, and a classification of organisms based on their biochemical machinery. The three levels of classification (from bonds to metabolic pathways) allowed to map and perceive chemical similarities between metabolic pathways even for pathways of different types of metabolism and pathways that do not share similarities in terms of EC numbers. Mapping of PES by neural networks (NNs). In a first series of experiments, ensembles of Feed-Forward NNs (EnsFFNNs) and Associative Neural Networks (ASNNs) were trained to reproduce PES represented by the Lennard-Jones (LJ) analytical potential function. The accuracy of the method was assessed by comparing the results of molecular dynamics simulations (thermal, structural, and dynamic properties) obtained from the NNs-PES and from the LJ function. The results indicated that for LJ-type potentials, NNs can be trained to generate accurate PES to be used in molecular simulations. EnsFFNNs and ASNNs gave better results than single FFNNs. A remarkable ability of the NNs models to interpolate between distant curves and accurately reproduce potentials to be used in molecular simulations is shown. The purpose of the first study was to systematically analyse the accuracy of different NNs. Our main motivation, however, is reflected in the next study: the mapping of multidimensional PES by NNs to simulate, by Molecular Dynamics or Monte Carlo, the adsorption and self-assembly of solvated organic molecules on noble-metal electrodes. Indeed, for such complex and heterogeneous systems the development of suitable analytical functions that fit quantum mechanical interaction energies is a non-trivial or even impossible task. The data consisted of energy values, from Density Functional Theory (DFT) calculations, at different distances, for several molecular orientations and three electrode adsorption sites. The results indicate that NNs require a data set large enough to cover well the diversity of possible interaction sites, distances, and orientations. NNs trained with such data sets can perform equally well or even better than analytical functions. Therefore, they can be used in molecular simulations, particularly for the ethanol/Au (111) interface which is the case studied in the present Thesis. Once properly trained, the networks are able to produce, as output, any required number of energy points for accurate interpolations.

Exposição profissional a chumbo : utilização da protoporfirina-zinco (PPZ) na vigilância de saúde de trabalhadores ex

Resumo Uma estratégia de avaliação e prevenção de riscos na exposição a agentes químicos deve ter sempre em conta que a vigilância do ambiente de trabalho e a da saúde dos trabalhadores são aspectos complementares de uma mesma realidade – os riscos resultantes da interacção entre um agente químico e os trabalhadores a ele expostos. Se à Vigilância Ambiental compete apreciar o risco, pela caracterização do agente no ambiente de trabalho, a Vigilância Biológica pronuncia-se sobre a interacção entre o tóxico e o organismo, avaliando a resposta à agressão química e a evolução das reacções de adaptação ou de desajuste face à absorção do tóxico. Os Indicadores Biológicos, deste modo, assumem um estatuto de instrumento privilegiado na vigilância da saúde dos trabalhadores expostos, na medida em que medem a quantidade de tóxico que efectivamente penetrou e foi absorvido, ou o resultado (efeito) determinado por essa mesma dose. O presente estudo procura contribuir para a definição de um quadro metodológico de utilização dos Indicadores Biológicos na avaliação/gestão da exposição profissional ao chumbo, designadamente apreciando a variação da protoporfirina-zinco (PPZ), indicador até ao presente ainda não utilizado em Portugal. O chumbo é um metal de ocorrência natural, cujos níveis nos diversos ecossistemas resultam, principalmente, das actividades antropogénicas de natureza doméstica e industrial. A sua capacidade poluente é assinalável, representando uma fonte de exposição permanente para o homem, demonstrável pela sua constante presença no organismo apesar de não desempenhar qualquer tipo de função fisiológica. São actualmente inúmeras as suas aplicações, tornando a exposição profissional ao chumbo uma realidade vasta: indústrias de acumuladores eléctricos, de vidros, de plásticos e de munições, construção civil, manutenção e reparação automóvel e de navios, fabrico de tintas, indústrias electrónicas, fundições e actividades de soldadura são, entre outras, situações onde é uma realidade a ter em conta. A penetração dos compostos inorgânicos de chumbo no organismo efectua-se principalmente por via respiratória, não sendo, no entanto, desprezável, a sua penetração por via digestiva. As partículas absorvidas são transportadas pelo sangue principalmente ligadas aos eritrocitos (95%), distribuem-se pelos tecidos moles e depositam-se essencialmente no tecido ósseo, onde representam mais de 90% da carga corporal do total absorvido e tendo aí um elevado tempo de semi-vida (mais de 20 anos). Não é metabolizado no organismo e a sua eliminação efectua-se essencialmente por via renal,sendo igualmente excretado, em menor escala, através das fezes, do suor, da saliva, das faneras e do leite materno. O conhecimento científico evidencia que concentrações sanguíneas de chumbo entre 20 e 50 mg/dL são susceptíveis de determinar efeitos adversos no homem, podendo ser afectados o sistema hematopoiético, o sistema nervoso, o sistema cardiovascular, o sistema reprodutor e o sistema imunitário. Contudo, ainda muito há a clarificar no âmbito da toxicidade do chumbo. Os níveis de exposição a que correspondem as alterações nos diversos órgãos e sistemas continuam a ser motivo de alguma controvérsia. As características carcinogénicas e mutagénicas do chumbo são, ainda, um campo de vasta exigência de investigação. A intoxicação por chumbo e seus sais (Saturnismo) de origem ocupacional é reconhecida em Portugal como doença profissional (grupo 1 - Doenças Provocadas por Agentes Químicos, da Lista das Doenças Profissionais). É uma intoxicação do tipo crónico, fruto da absorção contínua de doses relativamente pequenas durante longo período, evidenciando-se no seu início por sinais e sintomas vagos e difusos de grande inespecificidade, que podem incluir, nomeadamente, perda de apetite, sabor metálico na boca, palidez, mal-estar e fadiga, cefaleias, mialgias e artralgias, irritabilidade, tremores finos, obstipação, cólicas abdominais, insónias, déficit da memória de curto prazo e da capacidade de concentração. Um importante conjunto de indicadores biológicos pode ser utilizado na vigilância periódica da saúde de trabalhadores nestas condições de exposição. Tais indicadores (de dose ou de efeito), encerram diferentes significados e comportam distintas exigências, competindo ao Médico do Trabalho, no âmbito dos programas de prevenção dos efeitos adversos relacionados com a exposição profissional a chumbo, seleccionar a sua utilização e interpretar a sua informação, de modo a avaliar a interacção do tóxico com o organismo numa fase de reversibilidade. O presente estudo envolveu 180 trabalhadores dos quais 110 apresentavam plumbémias (Pb-S) iguais ou superiores a 40 mg/dL. Além da Pb-S, a todos foi doseada a protoporfirina-zinco (PPZ) e efectuado o Hemograma e a cerca de 25% foi determinada a concentração do ácido d-aminolevulínico urinário (ALA-U). Os doseamentos da PPZ efectuados em amostra de sangue capilar através de um hematofluorímetro portátil revelaram-se de total fiabilidade, dando significado a uma técnica de fácil execução e baixo custo. A avaliação do tipo de colheita urinária para doseamento do ALA-U concluiu pela necessidade de recurso a urinas de 24 horas.Os resultados do estudo evidenciaram uma elevada associação entre a PPZ e a Pb-S, com uma maior magnitude e de início mais precoce do que o que registado na associação da Pb-S com o ALA-U. Revelaram, ainda, fracos níveis de associação da hemoglobina (e outros parâmetros hematológicos) com a Pb-S. E demonstraram para um cut-off de 100 mg/ dL de PPZ, taxa de falsos negativos e falsos positivos, para plumbémias a partir de 70 mg/dL, inferiores a 20%. Assim, concluiu-se que, nos protocolos de vigilância de saúde de trabalhadores expostos a chumbo, o doseamento da PPZ por hematofluorímetro, em sangue de colheita capilar, é adequado, fiável e de realização preferencial em relação ao do ALA-U. Concluiu-se, também, que a realização do hemograma apenas se justifica em situações individuais que clinicamente o tornem aconselhável. E que estes protocolos devem incluir a realização da Pb-S e da PPZ, podendo, em situações de controlo rigoroso (ambiental, biológico e clínico), basear-se apenas na determinação da PPZ reservando os outros indicadores para aprofundar a investigação médica nos casos de taxas elevadas desta ou de situações limitantes. ■ Résumée Une stratégie d’évaluation et de prévention des risques d’exposition aux agents chimiques doit toujours tenir en considération que la vigilance du lieu de travail et de la santé des travailleurs sont des aspects complémentaires d’une même réalité – les risques résultant d’une interaction entre l’agent chimique et les travailleurs exposés. Si c’est à la Vigilance Ambiantale de juger le risque, par la caractérisation de l’agent dans le lieu de travail, la Vigilance Biologique, elle, se prononce sur l’interaction entre le toxique et l’organisme, évaluant la réponse à l’agression chimique et l’évolution des réactions d’adaptation ou de rupture face à l’absorption du toxique. Les Indicateurs Biologiques assument ainsi un statut d’instrument privilégié de vigilance de la santé des travailleurs exposés, dans la mesure où ils déterminent la quantité de toxique qui a effectivement été pénétré et absorbé, ou le résultat (effet) déterminé par cette dose. Cette étude-ci cherche à contribuer à la définition d’un cadre méthodologique d’utilisation des Indicateurs Biologiques dans l’évaluation/ gestion de l’exposition professionnelle au plomb inorganique, évaluant spécialement le comportement de la protoporphirine-zinc (PPZ), indicateur pas encore utilisé au Portugal.Le plomb est un métal d’occurrence naturelle dont les niveaux dans les différents écosystèmes en résultent, principalement, des activités anthropogéniques de nature domestique et industrielle. Sa capacité polluante peut être signalée, représentant une source d’exposition permanente pour l’homme, celle-ci démontrable par sa présence continue dans l’organisme, même si elle n’y accomplit aucune fonction physiologique. Actuellement ses applications sont innombrables, faisant de l’exposition professionnelle au plomb une réalité de grande ampleur : industries d’accumulateurs électriques, de verre, de plastique et de munitions, bâtiments, manutention et réparation automobile et navale, fabrication d’encres, industries électroniques, fontes et activités de soudure sont, entre autres, des situations réelles a en tenir compte. La pénétration du plomb inorganique dans l’organisme se fait principalement par voie respiratoire, pouvant se faire également par voie digestive. Les particules absorbées sont transportées par le sang, surtout liées aux érythrocytes (95%), se repartent à travers les tissus mous et se déposent essentiellement dans le tissu osseux, où elles représentent plus de 90% de la charge corporelle de ce qui a été absorbé et ont un temps de demi-vie élevé (plus de 20 ans). Le plomb n’est pas métabolisé dans l’organisme et son élimination se fait essentiellement par voie rénale, pouvant tout de même, à une moindre échelle, être excrété dans les fèces, de la sueur, de la salive, des ongles, des cheveux et du lait maternel. La connaissance scientifique met en évidence que des concentrations sanguines de plomb entre 20 et 50 mg/dL sont susceptibles de déterminer des effets adverses dans l’homme, pouvant les systèmes hématopoïétique, nerveux, cardiovasculaire, reproducteur et immunitaire en être affectés. Cependant, il en reste beaucoup à éclaircir dans le domaine de la toxicité du plomb. Les niveaux d’exposition auxquels correspondent les modifications des divers organes et systèmes, demeurent toujours sujet de quelque controverse. Les caractéristiques carcinogèniques et mutagèniques du plomb restent toujours un champ d’investigation d’une grande exigence. L’intoxication par le plomb et ses sels (Saturnisme) d’origine occupationnelle est reconnue, au Portugal, comme une maladie professionnelle (groupe 1- Maladies Provoquées par des Agents Chimiques, de la Liste des Maladies Professionnelles). C’est une intoxication du tipe chronique, due à l’absorption continue de doses relativement petites pendant une longue période, mise en évidence à travers des signes et des symptômes vagues et diffus sans grande spécificité, lesquels peuvent inclure, particulièrement, le manque d’appétit, goût métallique dans la bouche, pâleur, malaise et fatigue, céphalées, myalgies et arthralgies, irritabilité, tremblements fins, constipation, coliques abdominales, insomnies, déficit de la mémoire à court terme et de la capacité de concentration.Un ensemble important d’indicateurs biologiques peut être employé dans la vigilance périodique de la santé des travailleurs dans ces conditions d’exposition. Ces indicateurs (de dose ou d’effet) renferment différentes significations et comportent diverses exigences, devant le Médecin de Travail, dans le domaine des programmes de prévention des effets adverses qui sont en relation avec l’exposition professionnelle au plomb, sélectionner son utilisation et interpréter son information de façon à évaluer l’interaction de l’élément toxique avec l’organisme à un stade de réversibilité. L’étude ci-présent engloba 180 travailleurs desquels 110 présentaient des plombémies (Pb-S) égales ou supérieures à 40 mg/dL. À part la Pb-S, la protoporphyrine-zinc (PPZ) leur a été prise en dosage et un Hémogramme fut effectué et fut déterminé l’acide d- aminolévulinique urinaire (ALA-U) sur environ 25% des travailleurs. Le dosage de la PPZ efectué en échantillon de sang capillaire par un fluorimètre portable, s’est accomplit d’une fiabilité total, donnant du sgnificat à une téchnique de facile execution et bas prix. L’évaluation de la prise urinaire par dosage du ALA-U conclut au besoin d’un recours aux urines de 24 heures Les résultats de l’étude ont mis en évidence une association élevée entre la PPZ et la Pb- S, avec une intensité majeure et de début plus précoce par rapport à celui qui fut registré lors de l’association de la Pb-S avec la ALA-U. Ces résultats ont également montré de faibles niveaux d’association entre l’hémoglobine (et autres paramètres hématologiques) et la Pb-S. Ils ont démontré aussi, une valeur de cut-off de 100 mg/dL de PPZ, des taux de faux négatifs et faux positifs, pour des plombémies de 70 mg/dL, inférieurs à 20%. On peut donc conclure que dans les protocoles de vigilance de la santé des travailleurs exposés au plomb, le dosage de la PPZ par fluorimetrie dans le sang capillaire est adéquat, fiable et de réalisation préférentielle par rapport à celui du ALA-U. On peut également conclure que la réalisation de l’hémogramme ne se justifie que dans les cas individuels où, cliniquement, celui-ci est conseillé. De plus, ces protocoles doivent inclure la réalisation de la Pb-S et de la PPZ, pouvant, en cas de contrôle rigoureux (ambiantal, biologique et clinique), s’appuyer que dans la détermination de la PPZ réservant les autres indicateurs pour approfondir l’investigation médicale dans les cas où les taux de celle-ci sont élevés ou dans les cas de situations limitantes. ■ Summary Any strategy to evaluate and prevent the risks of chemical agents exposure must always regard the work environment and workers health as complementary aspects of one reality - the resulting risks from the interaction between the chemical agent and the exposed workers. It is the responsibility of Environmental Monitoring to evaluate the risks of exposure by the characterization of the chemical agent in the work environment. Biological Monitoring, on the other hand, pronounces itself over the toxin and body interaction, evaluating human response to the chemical aggression and the body adaptations to the toxic absorption. Biological Exposure Indices (BEI) assume, therefore, a privileged status among exposed workers' health monitoring instruments, as they measure the actual penetrated and absorbed toxic quantity and the effect it produces. This research study aims to contribute to the definition of a methodological strategy on the utilization of BEI’s in evaluating inorganic lead's occupational exposure, more specifically appreciating the zinc protoporphyrin (ZPP) variation, an index that has never been taken under consideration in Portugal until now. Lead is a natural metal whose ecosystem’s levels are mainly due to domestic and industrial anthropogenic activities. Its pollutant capacity is notable, representing a permanent exposure risk shown by its constant presence in the human body, although it has no physiologic function. Nowadays, lead's applications are countless, turning its professional exposure a huge reality: storage batteries industries, glass industries, plasterers and munitions industries, building construction, ships and motor car maintenance and repairing, ink manufacture, electronics industries, foundries and other soldering activities are, among so many other, realities to attend to. Respiration is the main cause of human body's inorganic lead absorption, although digestive pathway must not to be ignored. The absorbed particles are transported by blood, essentially bounded to erythrocytes (95%). It is distributed by soft tissues and settled mainly on bone tissues, where it represents approximately 90% of the total body charge and has a high half-life time (more than 20 years). It is not metabolized by the organism, its elimination being effectuated by renal activity and, in smaller scale, through lees, sweat, saliva, nails, hair and maternal milk.Scientific knowledge shows that concentrations of lead in blood between 20 e 50 mg/dL are susceptible to determine adverse effects in man and able to affect the hematopoietic system, the nervous system, the cardiovascular system, the reproductive system and the immunological system. Nevertheless, there's still much to be learned and clarified about lead's toxicity. The correlation between exposure levels and human's systems and organs alteration levels continues to be a centre of controversies. Still, lead's carcinogenic and mutagenic characteristics continue to be a high demanding research field. Intoxication by lead and its compounds (saturnism), from occupational origin, is recognized in Portugal as an occupational disease, included in Group 1 - Chemical Agents Caused Diseases, on the Occupational Diseases List. It is a chronic intoxication caused by a continuous absorption of small doses, throughout a long period of time. Its signs and symptoms are diffuse and imprecise, of great unspecificity, such as loss of appetite, metallic flavor in the mouth, paleness, ailment and fatigue, headaches, myalgia and arthralgia, irritability, thin tremors, constipation, abdominal pain, insomnias, short memory loses and inability to concentrate. A considered number of BEI’s can be used in Periodic Health Monitoring of workers in such exposure conditions. Such BEI (dose indices or effect indices) provide different meanings and imply different procedures, being Occupational Doctors responsibility, in the context of lead related adverse effects preventive programmes, to select and interpret its information, in order to evaluate the interaction between toxic and organism in a reversible phase of the toxic action. The present research study involved 180 workers, 110 of which presented blood lead levels (PbB) above or equal to 40 mg/dL. Besides PbB, all workers has been evaluated for zinc protoporphyrin levels (PPZ) and submitted to a haemogram. About 25% of the workers were selected for d-aminolevulinic urinary acid (ALA-U) determination. The evaluation of PPZ, by a portable hematofluorometer using capillary blood samples, turned out to be an easy procedure with low costs and total warrantability. As in regard for ALA- U procedure, it was concluded the necessity of 24 hours urine samples. This research results underlined a strong connection between ZPP and PbB, which was found to to be stringer and to begin earlier than it was registered for PbB and ALA-U association. The same study also revealed a low association level between PbB and hemoglobin or other hematological indices. It was also verified less than 20% of false negatives and false positives cases when admitted a ZPP 100 mg/dL cut off value for PbB³ 70 mg/dL. As in result it was concluded that in Health Monit

Problèmes posés par la population sporo-pollinique d'un niveau à plantes de la série de Buarcos (Portugal)

L'analyse pollinique des sédiments argileux, riches en macrorestes de végétaux, du "Belasien» (Crétacé inférieur) de Buarcos a permis reconnaître la présence d'une riche microflore. On a, nottament, reconnu la présence de: Concavisporites punctatus, Auritulinosporites complexis, Chomotriletes sp., Trilobosporites cf. bernissartensis, Apiculatisporites vulgaris, Cicatricosisporites sp. 1, Cicatricosisporites sp. 2. Cicatricosisporites sp. 3, Cicatricosisporites sp. 4, Cicatricosisporites sp. 5, Cicatricosisporites sp. 6. Costatoperforosporites fistulosus, Ceratosporites sp. l, C. cf. equalis, Ischyosporites teixeirae n. sp., Liburnisporites sp. AequitriraditeS cf. spinulosus, Cedripites lusitanicus, Clavatipollenites cf. hughesi. Après discution de la stratigraphie des diffèrentes espèces on sugère une âge barremo-aptien pour cette association malgré la présence de l'ensemble Apiculalisporites vulgaris-Classopollls major plutôt d'âge Aptien.