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.

Climate change impacts on portuguese energy system in 2050

Dissertação apresentada na Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa para a obtenção do grau de Mestre em Engenharia do Ambiente, perfil Gestão e Sistemas Ambientais

Société rurale, société urbaine: espaces en interaction

La sociologie et les nouveaux défis de la modernisation, Porto, pp. 315-326

Algorithms and methodologies for decision support in energy efficiency on buildings

Dissertação apresentada na faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa para a obtenção do grau de Mestre em Engenharia Electrotécnica e de Computadores

Simulation of the velocity field in compound channel flow using different closure models

1st European IAHR Congress,6-4 May, Edinburg, Scotland

Modelling a cable structure for a new wind energy production device

Dissertação apresentada na Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa para obtenção do grau de Mestre em Engenharia Mecânica Especialização em Concepção e Produção

Energy-dispersive x-ray fluorescence analysis of modern coloured glasses from Marinha Grande (Portugal)

X-Ray Spectrom. 2003; 32: 396–401

Energy transduction by respiratory complex I

Dissertation presented to obtain a PhD degree in Biochemistry at the Instituto de Tecnologia Química e Biológica, Universidade Nova de Lisboa

Electron driven reactions in complexes embedded in superfluid helium droplets

A thesis submitted to the University of Innsbruck for the doctor degree in Natural Sciences, Physics and New University of Lisbon for the doctor degree in Physics, Atomic and Molecular Physics

Dynamic and interaction of cytochrome c with Pf1 virus

Dissertação apresentada na Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa para obtenção do grau de Mestre em BioOrgânica

Industry level foresight: designing foresight methods for Lithuanian energy sector

Enterprise and Work Innovation Studies,6,IET, pp.9-51

Projecto de um circuito conversor DC-DC para aplicações de “energy harvest”

Dissertação apresentada na Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa para a obtenção do grau de Mestre em Engenharia Electrotécnica e de Computadores

SOLAP+: extending the interaction model

Thesis submitted to Faculdade de Ciências e Tecnologia of the Universidade Nova de Lisboa, in partial fulfillment of the requirements for the degree of Master in Computer Science

Calcificações vasculares nos doentes em diálise : elo de ligação entre doença óssea e doença vascular

RESUMO: A presente dissertação para tese de doutoramento apresenta o desenvolvimento e a validação de um método simples e original para o diagnóstico de calcificações vasculares em doentes em diálise, utilizando um score semiquantitativo criado por nós e obtido em RX simples da bacia e das mãos, denominado score de calcifi cação vascular simples. Demonstramos que este score vascular simples é preditor de risco cardiovascular nos doentes em diálise. O score de calcificação vascular simples associou-se ainda à baixa densidade mineral óssea avaliada por dual energy X -ray absortiometry (DXA) no colo do fémur. Verifi camos igualmente que, em doentes em diálise, as calcifi cações coronárias quantifi cadas pelo score de Agatston e o score de calcifi cação vascular simples se associaram a um menor volume ósseo avaliado em biopsias ósseas. Estes trabalhos corroboram a hipótese da existência de um elo de ligação entre a doença óssea e a doença vascular nos doentes em diálise, e um dos elementos que contribuem para este elo de ligação podem ser as calcificações vasculares. Este score de calcificação vascular simples avalia calcifi cações em artérias de grande, médio e pequeno calibre, e inclui os dois padrões radiológicos de calcificação: calcificação linear, associada à calcifi cação da camada média da parede arterial, e calcificação irregular, associada à calcifi cação da camada íntima arterial1. Nos diferentes trabalhos por nós publicados demonstramos que as calcificações vasculares avaliadas por este método simples e barato permitem a identificação de indivíduos com elevado risco cardiovascular. Este score vascular associa -se a maior risco de mortalidade cardiovascular2, de mortalidade de causa global3, de internamentos cardiovasculares2, de doença ardiovascular2, de doença arterial periférica2,4,de calcifi cações valvulares5 e de rigidez arterial3. As guidelines KDIGO (Kidney disease: improving global outcomes), publicadas em 2009,sugerem que os doentes renais crónicos nos estadios 3 a 5, com calcificações vasculares e valvulares, devem ser considerados como apresentando o mais elevado risco cardiovascular6. A elevada mortalidade dos doentes renais crónicos não é totalmente explicada pelos fatores de risco tradicionais7. A organização KDIGO defende, desde 2006, a hipótese da existência de um elo de ligação entre a doença óssea e a doença vascular8. Esta ligação pode ser explicada pelas alterações do metabolismo mineral e ósseo e pela sua interação com as calcificações vasculares. Verificamos, nos nossos trabalhos, uma associação entre calcifi cações vasculares e doença óssea. O baixo volume ósseo diagnosticado por análise histomorfométrica de biopsias ósseas foi preditor de maior risco de calcificações vasculares avaliadas pelo score de calcifi cação vascular simples (dados apresentados nesta dissertação, no capítulo 6) e pelo score coronário de Agatston num grupo de doentes em diálise9. A contribuição original deste artigo9 foi considerada merecedora de um editorial feito pelo Dr. Gérard London10, investigador líder na área da calcificação vascular dos doentes renais crónicos e actual Presidente da EDTA (European Dialysis and Transplantation Association). Fomos também os primeiros a descrever uma associação independente e inversa entre a densidade mineral avaliada no colo do fémur por DXA (dual energy X -ray absortiometry) com calcificações vasculares avaliadas pelo score de calcificação vascular simples, com rigidez arterial avaliada por velocidade de onda de pulsocarotidofemoral e com doença arterial periférica diagnosticada por critérios clínicos11. Fomos igualmente os primeiros a mostrar uma correlação signifi cativa entre a densidade mineral óssea avaliada por DXA no colo do fémur, mas não na coluna lombar, com a espessura cortical avaliada por análise histomorfométrica em biopsia óssea12. O nosso estudo atribui pela primeira vez à DXA um papel no diagnóstico de porosidade cortical nos doentes em diálise. A utilidade da avaliação diferencial da densidade mineral óssea cortical e trabecular necessita ainda de ser confirmada em estudos prospectivos. Este achado inovador do nosso estudo foi mencionado pela ERBP (European Renal Best Practice) no comentário feito à posição da KDIGO que considera ser reduzida a utilidade da densidade mineral óssea nos doentes em diálise13. Dois dos trabalhos incluídos nesta dissertação foram referenciados nas guidelines KDIGO 2009 para avaliar a prevalência das calcificações vasculares (KDIGO 2009: Tabela suplementar 10, Fig. 3.6) e para validar a associação entre calcificações vasculares e mortalidade cardiovascular (KDIGO 2009: Tabela suplementar 12, Fig. 3.7)6. A inclusão destes nossos dois estudos nas referências destas guidelines, que utilizaram o exigente sistema GRADE (Grades of recommendation, assessment, development, and evaluation) na classificação e selecção dos estudos, valida o interesse científico dos nossos trabalhos. O diagnóstico de calcificações vasculares tem um interesse prático para os doentes renais crónicos. A presença de calcifi cações vasculares é um sinal de alerta para a existência de um elevado risco cardiovascular, e esta informação pode ser utilizada para modificar a terapêutica nestes doentes6. Diferentes métodos podem ser usados para diagnosticar calcificações vasculares nos doentes em diálise14,15. O score de calcificação vascular simples tem a vantagem da simplicidade e de poder ser facilmente interpretado pelo nefrologista, sem necessidade de um radiologista. A reprodutibilidade deste score já foi demonstrada por diferentes grupos em estudos nacionais e internacionais16-24. Nestes estudos foi demonstrado que as calcifi cações vasculares avaliadas pelo método criado por nós são preditoras de maior risco de eventos cardiovasculares16, de amputações dos membros inferiores17, de velocidade de onda de pulso18,19, de calcificações corneanas e conjuntivais20 e de calcifi cações coronárias21. Também foi demonstrada uma associação inversa entre o score de calcificação vascular simples com os níveis séricos de PTH21, com os níveis de 25(OH)vitamina D 22,23 e com os níveis de fetuína A19,24. Todos estes estudos, realizados por diferentes grupos, que utilizaram o score de calcificação vascular simples na sua metodologia, comprovam a facilidade de utilização deste score e a concordância de resultados atestam a sua reprodutibilidade e a utilidade na avaliação dos doentes renais crónicos. ---------------------------ABSTRACT: This thesis presents the development and validation of a simple and original method to identify vascular calcifications in dialysis patients, using a semi -quantitative score that we have created and that is obtained in plain X -ray of pelvis and hands. This score was named in different publications as “simple vascular calcifi cation score”. We have demonstrated that this score is a predictor of higher cardiovascular risk in dialysis patients. The simple vascular calcification score was also associated with lower mineral bone density evaluated by DXA in femoral neck. In hemodialysis patients coronary calcifications evaluated by the coronary Agatston score and by the simple vascular calcification score were associated with lower bone volume analysed in bone biopsies. These studies corroborate the hypothesis of the existence of a link between bone disease and vascular disease in dialysis patients and one of the elements of this link may be vascular calcifications. This simple vascular calcification score identifi es calcifications in large, medium and small calibre arteries and includes the two radiological patterns of arterial calcifi cation: linear calcification which has been associated with the calcifi cation of the media layer of the arterial wall and irregular and patchy calcification which has been associated with the calcifi cation of the intima layer of the arterial wall1. In the several studies that we have published we have demonstrated that vascular calcifications evaluated by this simple and inexpensive method allow the identification of patients with high cardiovascular risk. This simple vascular calcification score is an independent predictor of cardiovascular mortality2, all -cause mortality3, cardiovascular hospitalizations2, cardiovascular disease2, peripheral artery disease2,4, valvular calcifi cations5 and arterial stiffness3.KDIGO (Kidney Disease: Improving Global Outcomes) guidelines published in 2009 suggest that chronic kidney disease patients in stages 3 to 5, with vascular and valvular calcifications should be considered to be at the highest cardiovascular risk6. The high mortality of chronic kidney disease patients is not completely explained by the traditional risk factors7 and KDIGO group supports, since 2006, the hypothesis of the existence of a link between bone disease and vascular disease8.This link may be explained by the alterations of the bone and mineral metabolism and their interaction with development and progression of vascular calcifications. We have also verifi ed in our studies the existence of an association between vascular calcifications and bone disease. Low bone volume diagnosed by histomorphometric analysis of bone biopsies, in a group of dialysis patients, was independently associated with the simple vascular calcification score (data presented in this thesis,chapter 6) and with coronary calcifications evaluated by the Agatston score9. The original contribution of this article published in CJASN9 deserved a commentary in an Editorial written by Prof. Gérard London10 leader investigator in this area and current EDTA (European Dialysis and Transplantation Association) President. We were also the fi rst group to describe an independent and inverse association between bone mineral density evaluated in the femoral neck by DXA (dual energy X -ray absortiometry) with vascular calcifications evaluated by the simple vascular calcification score, with arterial stiffness evaluated by carotid-femoral pulse wave velocity and with peripheral artery disease diagnosed by clinical criteria11. We were also the first group to demonstrate a significant correlation between bone mineral density evaluated by DXA in femoral neck but not in lumbar spine, with cortical thickness evaluated by histomorphometric analysis of bone biopsy12. Our study has attributed to DXA, for the first time, a role in the diagnosis of cortical porosity in dialysis patients. The clinical utility of the differential evaluation of bone mineral density in cortical or trabecular bone needs, however, to be confi rmed in prospective studies. This original fi nding of our study was mentioned by ERBP (European Renal Best Practice) commenting the KDIGO position in relation with the reduced utility of bone mineral density evaluation in dialysis patients13. Two of the studies included in this thesis have been integrated in a group of studies selected as references by the KDIGO guidelines published in 2009 to evaluate the prevalence of vascular calcifications in CKD patients (KDIGO 2009: Supplementary Table 10, Fig. 3.6) and to corroborate the association between vascular calcifications and cardiovascular mortality (KDIGO 2009: Supplementary Table 12, Fig. 3.7)6. The inclusion of both studies as references in the KDIGO guidelines that have used the exigent GRADE system (Grades of Recommendation, Assessment, Development, and Evaluation) in the classifi cation and selection of studies, validates the scientifi c value of our studies. The diagnosis of vascular calcifi cations has a practical interest for chronic kidney disease patients. The presence of vascular calcifications is an alert sign to the existence of a high cardiovascular risk and this information may be used to modify the treatment of these patients6. Different methods may be used to detect the presence of vascular calcifications in dialysis patients14,15. The simple vascular calcifi cation score has the advantage of being simple, inexpensive and easily evaluated by the Nephrologist without the need for a Radiologist interpretation. The reproducibility of this method has already been demonstrated by other groups in national and international studies16 -24. It was demonstrated in those studies that vascular calcifi cations evaluated by the method created by us, predict higher risk of cardiovascular events16, higher risk of lower limbs amputations17, higher pulse wave velocity18,19, corneal and conjuntival calcifi cations 20 and coronary calcifi cations21. A negative association between the simple vascular calcification score and PTH levels21, 25(OH) vitamin D levels22,23 and Fetuin A levels19,24 has also been demonstrated. All these studies performed by different groups that have used the simple vascular calcifi cation score in their methods demonstrate that this score is simple, useful and reproducible in the evaluation of chronic kidney disease patients simple, useful and reproducible in the evaluation of chronic kidney disease patients.

Avaliação da comunicação da sustentabilidade de energy utilities e análise dos principais rankings de sustentabilidade

Dissertação para obtenção do Grau de Mestre em Engenharia do Ambiente Perfil de Gestão e Sistemas Ambientais