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.

Synthesis, characterization and reactivity of new cobalt, palladium and nickel complexes bearing α-diimines and P, O ligands

This work describes the synthesis and characterization of a series of new α-diimine and P,O, β-keto and acetamide phosphines ligands, and their complexation to Ni(II), Co(II),Co(III) and Pd(II) to obtain a series of new compounds aiming to study their structural characteristics and to test their catalytic activity. All the compounds synthesized were characterized by the usual spectroscopic and spectrometric techniques: Elemental Analysis, MALDI-TOF-MS spectrometry, IR, UV-vis, 1H, 13C and 31P NMR spectroscopies. Some of the paramagnetic compounds were also characterized by EPR. For the majority of the compounds it was possible to solve their solid state structure by single crystal X-ray diffraction. Tests for olefin polymerization were performed in order to determine the catalytic activity of the Co(II) complexes. Chapter I presents a brief introduction to homogenous catalysis, highlighting the reactions catalyzed by the type of compounds described in this thesis, namely olefin polymerization and oligomerization and reactions catalyzed by the complexes bearing α-diimines and P,O type ligands. Chapter II is dedicated to the description of the synthesis of new α-diimines cobalt (II) complexes, of general formula [CoX2(α-diimine)], where X = Cl or I and the α-diimines are bis(aryl)acenaphthenequinonediimine) (Ar-BIAN) and 1,4-diaryl-2,3-dimethyl-1,4-diaza-1,3-butadiene (Ar-DAB). Structures solved by single crystal X-ray diffraction were obtained for all the described complexes. For some of the compounds, X-band EPR measurements were performed on polycrystalline samples, showing a high-spin Co(II) (S = 3/2) ion, in a distorted axial environment. EPR single crystal experiments on two of the compounds allowed us to determine the g tensor orientation in the molecular structure. In Chapter III we continue with the synthesis and characterization of more cobalt (II)complexes bearing α-diimines of general formula [CoX2(α-diimine)], with X = Cl or I and α-diimines are bis(aryl)acenaphthenequinonediimine) (Ar-BIAN) and 1,4-diaryl-2,3-dimethyl- 1,4-diaza-1,3-butadiene (Ar-DAB). The structures of three of the new compounds synthesized were determined by single crystal X-ray diffraction. A NMR paramagnetic characterization of all the compounds described is presented. Ethylene polymerization tests were done to determine the catalytic activity of several of the Co(II) complexes described in Chapter II and III and their results are shown. In Chapter IV a new rigid bidentate ligand, bis(1-naphthylimino)acenaphthene, and its complexes with Zn(II) and Pd(II), were synthesized. Both the ligand and its complexes show syn and anti isomers. Structures of the ligand and the anti isomer of the Pd(II) complex were solved by single crystal X-ray diffraction. All the compounds were characterized by elemental analysis, MALDI-TOF-MS spectrometry, and by IR, UV-vis, 1H, 13C, 1H-1H COSY, 1H-13C HSQC, 1H-13C HSQC-TOCSY and 1H-1H NOESY NMR when necessary. DFT studies showed that both conformers of [PdCl2(BIAN)] are isoenergetics and can be obtain experimentally. However, we can predict that the isomerization process is not available in square-planar complex, but is possible for the free ligand. The molecular geometry is very similar in both isomers, and only different orientations for naphthyl groups can be expected. Chapter V describes the synthesis of new P, O type ligands, β-keto phosphine, R2PCH2C(O)Ph, and acetamide phosphine R2PNHC(O)Me, as well as a series of new cobalt(III) complexes namely [(η5-C5H5)CoI2{Ph2PCH2C(O)Ph}], and [(η5- C5H5)CoI2{Ph2PNHC(O)Me}]. Treating these Co(III) compounds with an excess of Et3N, resulted in complexes η2-phosphinoenolate [(η5-C5H5)CoI{Ph2PCH…C(…O)Ph}] and η2- acetamide phosphine [(η5-C5H5)CoI{Ph2PN…C(…O)Me}]. Nickel (II) complexes were also obtained: cis-[Ni(Ph2PN…C(…O)Me)2] and cis-[Ni((i-Pr)2PN…C(…O)Me)2]. Their geometry and isomerism were discussed. Seven structures of the compounds described in this chapter were determined by single crystal X-ray diffraction. The general conclusions of this work can be found in Chapter VI.

Periplasmic nitrate reductases: structural and spectroscopic studies

Dissertação apresentada para obtenção do grau de Doutor em Bioquímica, especialidade Bioquímica-Física, pela Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa

Wrong sign and symmetric limits and non-decoupling in 2HDMs

We analyse the possibility that, in two Higgs doublet models, one or more of the Higgs couplings to fermions or to gauge bosons change sign, relative to the respective Higgs Standard Model couplings. Possible sign changes in the coupling of a neutral scalar to charged ones are also discussed. These wrong signs can have important physical consequences, manifesting themselves in Higgs production via gluon fusion or Higgs decay into two gluons or into two photons. We consider all possible wrong sign scenarios, and also the symmetric limit, in all possible Yukawa implementations of the two Higgs doublet model, in two different possibilities: the observed Higgs boson is the lightest CP-even scalar, or the heaviest one. We also analyse thoroughly the impact of the currently available LHC data on such scenarios. With all 8 TeV data analysed, all wrong sign scenarios are allowed in all Yukawa types, even at the 1 sigma level. However, we will show that B-physics constraints are crucial in excluding the possibility of wrong sign scenarios in the case where tan beta is below 1. We will also discuss the future prospects for probing the wrong sign scenarios at the next LHC run. Finally we will present a scenario where the alignment limit could be excluded due to non-decoupling in the case where the heavy CP-even Higgs is the one discovered at the LHC.

Assessment of heavy metals in the surroundings of coal-fired power plant

This paper aims to survey metal concentrations in soils in the vicinity of a coal-firedpower plant located in southwest of Portugal. Two annual sampling campaigns were carried out to measure a hypothetical soil contamination around the coal plant. The sampling area was divided into two subareas, both centered in the emission source, delimited by two concentric circles with radius of 6 km and 20 km. About 40 samplings points were defined in the influence area. Metals measurements were performed with a portable analytical X-ray dispersive energy fluorescence spectrometer identifying about 20 different elements in each sampling point. The most relevant elements measured included As, Cu, Fe, Hg, Pb, Ti and Zn in both sampling areas. Considering the results obtained in the first sampling campaign, arsenic is predominantly higher within the 6-20 km sampling area. The second sampling campaign showed that both sampling areas presented relatively similar metal concentrations except for Fe, Mn, Sr and Zn which concentration is higher within the 6-20 km sampling area. Also, As, Fe, Mn and Ti concentrations decreased significantly from the first to the second sampling campaign and their concentration were predominately higher in the NE-E and E-SE directions.

Catalytic activity of a benzoyl hydrazone based dimeric dicopper(II) complex in catechol and alcohol oxidation reactions

The benzoyl hydrazone based dimeric dicopper(II) complex [Cu2(R)(CH3O)(NO3)]2(CH3O)2 (R-Cu2+), recently reported by us, catalyzes the aerobic oxidation of catechols (catechol (S1), 3,5- itertiarybutylcatechol (S2) and 3-nitrocatechol (S3)) to the corresponding quinones (catecholase like activity), as shown by UV–Vis absorption spectroscopy in methanol/HEPES buffer (pH 8.2) medium at 25 C. The highest activity is observed for the substituted catechol (S2) with the electron donor tertiary butyl group, resulting in a turnover frequency (TOF) value of 1.13 103 h1. The complex R-Cu2+ also exhibits a good catalytic activity in the oxidation (without added solvent) of 1-phenylethanol to acetophenone by But OOH under low power (10 W) microwave (MW) irradiation. 2014 Elsevier B.V. All rights reserved.

Caracterização de fibras de ion jelly® obtidas por electrofiação

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

Robot visual localization through local feature fusion: an evaluation of multiple classifiers combination approaches

In the last decade, local image features have been widely used in robot visual localization. In order to assess image similarity, a strategy exploiting these features compares raw descriptors extracted from the current image with those in the models of places. This paper addresses the ensuing step in this process, where a combining function must be used to aggregate results and assign each place a score. Casting the problem in the multiple classifier systems framework, in this paper we compare several candidate combiners with respect to their performance in the visual localization task. For this evaluation, we selected the most popular methods in the class of non-trained combiners, namely the sum rule and product rule. A deeper insight into the potential of these combiners is provided through a discriminativity analysis involving the algebraic rules and two extensions of these methods: the threshold, as well as the weighted modifications. In addition, a voting method, previously used in robot visual localization, is assessed. Furthermore, we address the process of constructing a model of the environment by describing how the model granularity impacts upon performance. All combiners are tested on a visual localization task, carried out on a public dataset. It is experimentally demonstrated that the sum rule extensions globally achieve the best performance, confirming the general agreement on the robustness of this rule in other classification problems. The voting method, whilst competitive with the product rule in its standard form, is shown to be outperformed by its modified versions.

Person Localization Using Sensor Information Fusion

Nowadays the incredible grow of mobile devices market led to the need for location-aware applications. However, sometimes person location is difficult to obtain, since most of these devices only have a GPS (Global Positioning System) chip to retrieve location. In order to suppress this limitation and to provide location everywhere (even where a structured environment doesn’t exist) a wearable inertial navigation system is proposed, which is a convenient way to track people in situations where other localization systems fail. The system combines pedestrian dead reckoning with GPS, using widely available, low-cost and low-power hardware components. The system innovation is the information fusion and the use of probabilistic methods to learn persons gait behavior to correct, in real-time, the drift errors given by the sensors.

Localization system for pedestrians based on sensor and information fusion

Nowadays there is an increase of location-aware mobile applications. However, these applications only retrieve location with a mobile device's GPS chip. This means that in indoor or in more dense environments these applications don't work properly. To provide location information everywhere a pedestrian Inertial Navigation System (INS) is typically used, but these systems can have a large estimation error since, in order to turn the system wearable, they use low-cost and low-power sensors. In this work a pedestrian INS is proposed, where force sensors were included to combine with the accelerometer data in order to have a better detection of the stance phase of the human gait cycle, which leads to improvements in location estimation. Besides sensor fusion an information fusion architecture is proposed, based on the information from GPS and several inertial units placed on the pedestrian body, that will be used to learn the pedestrian gait behavior to correct, in real-time, the inertial sensors errors, thus improving location estimation.

Metal ions and protein folding: conformational and functional interplay

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

Cross-Reactions between Toxocara canis and Ascaris suum in the diagnosis of visceral larva migrans by western blotting technique

Visceral larva migrans (VLM) is a clinical syndrome caused by infection of man by Toxocara spp, the common roundworm of dogs and cats. Tissue migration of larval stages causes illness specially in children. Because larvae are difficult to detect in tissues, diagnosis is mostly based on serology. After the introduction of the enzyme-linked immunosorbent assay (ELISA) using the larval excretory-secretory antigen of T. canis (TES), the diagnosis specificity was greatly improved although cross-reactivity with other helminths are still being reported. In Brazil, diagnosis is routinely made after absorption of serum samples with Ascaris suum antigens, a nematode antigenicaly related with Ascaris lumbricoides which is a common intestinal nematode of children. In order to identify T. canis antigens that cross react to A. suum antigens we analyzed TES antigen by SDS-PAGE and Western blotting techniques. When we used serum samples from patients suspected of VLM and positive result by ELISA as well as a reference serum sample numerous bands were seen (molecular weight of 210-200 kDa, 116-97 kDa, 55-50 kDa and 35-29 kDa). Among these there is at least one band with molecular weight around 55-66 kDa that seem to be responsible for the cross-reactivity between T. canis e A. suum once it disappears when previous absorption of serum samples with A. suum antigens is performed

Development of electrospun Ion jelly fibers® for drug delivery

Dissertação para obtenção do Grau de Mestre em Biotecnologia

Assessing and comparing the total antioxidant capacity of commercial beverages: application to beers, wines, waters and soft drinks using TRAP, TEAC and FRAP methods.

This work measures and tries to compare the Antioxidant Capacity (AC) of 50 commercial beverages of different kinds: 6 wines, 12 beers, 18 soft drinks and 14 flavoured waters. Because there is no reference procedure established for this purpose, three different optical methods were used to analyse these samples: Total Radical trapping Antioxidant Parameter (TRAP), Trolox Equivalent Antioxidant Capacity (TEAC) and Ferric ion Reducing Antioxidant Parameter (FRAP). These methods differ on the chemical background and nature of redox system. The TRAP method involves the transfer of hydrogen atoms while TEAC and FRAP involves electron transfer reactions. The AC was also assessed against three antioxidants of reference, Ascorbic acid (AA), Gallic acid (GA) and 6-hydroxy-2,5,7,8-tetramethyl- 2-carboxylic acid (Trolox). The results obtained were analyzed statistically. Anova one-way tests were applied to all results and suggested that methods and standards exhibited significant statistical differences. The possible effect of sample features in the AC, such as gas, flavours, food colouring, sweeteners, acidity regulators, preservatives, stabilizers, vitamins, juice percentage, alcohol percentage, antioxidants and the colour was also investigated. The AC levels seemed to change with brand, kind of antioxidants added, and kind of flavour, depending on the sample. In general, higher ACs were obtained for FRAP as method, and beer for kind of sample, and the standard expressing the smaller AC values was GA.