Ontology-based access to sensor data streams

Sensor networks are increasingly becoming one of the main sources of Big Data on the Web. However, the observations that they produce are made available with heterogeneous schemas, vocabularies and data formats, making it difficult to share and reuse these data for other purposes than those for which they were originally set up. In this thesis we address these challenges, considering how we can transform streaming raw data to rich ontology-based information that is accessible through continuous queries for streaming data. Our main contribution is an ontology-based approach for providing data access and query capabilities to streaming data sources, allowing users to express their needs at a conceptual level, independent of implementation and language-specific details. We introduce novel query rewriting and data translation techniques that rely on mapping definitions relating streaming data models to ontological concepts. Specific contributions include: • The syntax and semantics of the SPARQLStream query language for ontologybased data access, and a query rewriting approach for transforming SPARQLStream queries into streaming algebra expressions. • The design of an ontology-based streaming data access engine that can internally reuse an existing data stream engine, complex event processor or sensor middleware, using R2RML mappings for defining relationships between streaming data models and ontology concepts. Concerning the sensor metadata of such streaming data sources, we have investigated how we can use raw measurements to characterize streaming data, producing enriched data descriptions in terms of ontological models. Our specific contributions are: • A representation of sensor data time series that captures gradient information that is useful to characterize types of sensor data. • A method for classifying sensor data time series and determining the type of data, using data mining techniques, and a method for extracting semantic sensor metadata features from the time series.

LiDAR remote sensing applied to forest resources assessment

Disponer de información precisa y actualizada de inventario forestal es una pieza clave para mejorar la gestión forestal sostenible y para proponer y evaluar políticas de conservación de bosques que permitan la reducción de emisiones de carbono debidas a la deforestación y degradación forestal (REDD). En este sentido, la tecnología LiDAR ha demostrado ser una herramienta perfecta para caracterizar y estimar de forma continua y en áreas extensas la estructura del bosque y las principales variables de inventario forestal. Variables como la biomasa, el número de pies, el volumen de madera, la altura dominante, el diámetro o la altura media son estimadas con una calidad comparable a los inventarios tradicionales de campo. La presente tesis se centra en analizar la aplicación de los denominados métodos de masa de inventario forestal con datos LIDAR bajo diferentes condiciones y características de masa forestal (bosque templados puros y mixtos) y utilizando diferentes bases de datos LiDAR (información proveniente de vuelo nacionales e información capturada de forma específica). Como consecuencia de lo anterior, se profundiza en la generación de inventarios forestales continuos con LiDAR en grandes áreas. Los métodos de masa se basan en la búsqueda de relaciones estadísticas entre variables predictoras derivadas de la nube de puntos LiDAR y las variables de inventario forestal medidas en campo con el objeto de generar una cartografía continua de inventario forestal. El rápido desarrollo de esta tecnología en los últimos años ha llevado a muchos países a implantar programas nacionales de captura de información LiDAR aerotransportada. Estos vuelos nacionales no están pensados ni diseñados para fines forestales por lo que es necesaria la evaluación de la validez de esta información LiDAR para la descripción de la estructura del bosque y la medición de variables forestales. Esta información podría suponer una drástica reducción de costes en la generación de información continua de alta resolución de inventario forestal. En el capítulo 2 se evalúa la estimación de variables forestales a partir de la información LiDAR capturada en el marco del Plan Nacional de Ortofotografía Aérea (PNOA-LiDAR) en España. Para ello se compara un vuelo específico diseñado para inventario forestal con la información de la misma zona capturada dentro del PNOA-LiDAR. El caso de estudio muestra cómo el ángulo de escaneo, la pendiente y orientación del terreno afectan de forma estadísticamente significativa, aunque con pequeñas diferencias, a la estimación de biomasa y variables de estructura forestal derivadas del LiDAR. La cobertura de copas resultó más afectada por estos factores que los percentiles de alturas. Considerando toda la zona de estudio, la estimación de la biomasa con ambas bases de datos no presentó diferencias estadísticamente significativas. Las simulaciones realizadas muestran que las diferencias medias en la estimación de biomasa entre un vuelo específico y el vuelo nacional podrán superar el 4% en áreas abruptas, con ángulos de escaneo altos y cuando la pendiente de la ladera no esté orientada hacia la línea de escaneo. En el capítulo 3 se desarrolla un estudio en masas mixtas y puras de pino silvestre y haya, con un enfoque multi-fuente empleando toda la información disponible (vuelos LiDAR nacionales de baja densidad de puntos, imágenes satelitales Landsat y parcelas permanentes del inventario forestal nacional español). Se concluye que este enfoque multi-fuente es adecuado para realizar inventarios forestales continuos de alta resolución en grandes superficies. Los errores obtenidos en la fase de ajuste y de validación de los modelos de área basimétrica y volumen son similares a los registrados por otros autores (usando un vuelo específico y parcelas de campo específicas). Se observan errores mayores en la variable número de pies que los encontrados en la literatura, que pueden ser explicados por la influencia de la metodología de parcelas de radio variable en esta variable. En los capítulos 4 y 5 se evalúan los métodos de masa para estimar biomasa y densidad de carbono en bosques tropicales. Para ello se trabaja con datos del Parque Nacional Volcán Poás (Costa Rica) en dos situaciones diferentes: i) se dispone de una cobertura completa LiDAR del área de estudio (capitulo 4) y ii) la cobertura LiDAR completa no es técnica o económicamente posible y se combina una cobertura incompleta de LiDAR con imágenes Landsat e información auxiliar para la estimación de biomasa y carbono (capitulo 5). En el capítulo 4 se valida un modelo LiDAR general de estimación de biomasa aérea en bosques tropicales y se compara con los resultados obtenidos con un modelo ajustado de forma específica para el área de estudio. Ambos modelos están basados en la variable altura media de copas (TCH por sus siglas en inglés) derivada del modelo digital LiDAR de altura de la vegetación. Los resultados en el área de estudio muestran que el modelo general es una alternativa fiable al ajuste de modelos específicos y que la biomasa aérea puede ser estimada en una nueva zona midiendo en campo únicamente la variable área basimétrica (BA). Para mejorar la aplicación de esta metodología es necesario definir en futuros trabajos procedimientos adecuados de medición de la variable área basimétrica en campo (localización, tamaño y forma de las parcelas de campo). La relación entre la altura media de copas del LiDAR y el área basimétrica (Coeficiente de Stock) obtenida en el área de estudio varía localmente. Por tanto es necesario contar con más información de campo para caracterizar la variabilidad del Coeficiente de Stock entre zonas de vida y si estrategias como la estratificación pueden reducir los errores en la estimación de biomasa y carbono en bosques tropicales. En el capítulo 5 se concluye que la combinación de una muestra sistemática de información LiDAR con una cobertura completa de imagen satelital de moderada resolución (e información auxiliar) es una alternativa efectiva para la realización de inventarios continuos en bosques tropicales. Esta metodología permite estimar altura de la vegetación, biomasa y carbono en grandes zonas donde la captura de una cobertura completa de LiDAR y la realización de un gran volumen de trabajo de campo es económica o/y técnicamente inviable. Las alternativas examinadas para la predicción de biomasa a partir de imágenes Landsat muestran una ligera disminución del coeficiente de determinación y un pequeño aumento del RMSE cuando la cobertura de LiDAR es reducida de forma considerable. Los resultados indican que la altura de la vegetación, la biomasa y la densidad de carbono pueden ser estimadas en bosques tropicales de forma adecuada usando coberturas de LIDAR bajas (entre el 5% y el 20% del área de estudio). ABSTRACT The availability of accurate and updated forest data is essential for improving sustainable forest management, promoting forest conservation policies and reducing carbon emissions from deforestation and forest degradation (REDD). In this sense, LiDAR technology proves to be a clear-cut tool for characterizing forest structure in large areas and assessing main forest-stand variables. Forest variables such as biomass, stem volume, basal area, mean diameter, mean height, dominant height, and stem number can be thus predicted with better or comparable quality than with costly traditional field inventories. In this thesis, it is analysed the potential of LiDAR technology for the estimation of plot-level forest variables under a range of conditions (conifer & broadleaf temperate forests and tropical forests) and different LiDAR capture characteristics (nationwide LiDAR information vs. specific forest LiDAR data). This study evaluates the application of LiDAR-based plot-level methods in large areas. These methods are based on statistical relationships between predictor variables (derived from airborne data) and field-measured variables to generate wall to wall forest inventories. The fast development of this technology in recent years has led to an increasing availability of national LiDAR datasets, usually developed for multiple purposes throughout an expanding number of countries and regions. The evaluation of the validity of nationwide LiDAR databases (not designed specifically for forest purposes) is needed and presents a great opportunity for substantially reducing the costs of forest inventories. In chapter 2, the suitability of Spanish nationwide LiDAR flight (PNOA) to estimate forest variables is analyzed and compared to a specifically forest designed LiDAR flight. This study case shows that scan angle, terrain slope and aspect significantly affect the assessment of most of the LiDAR-derived forest variables and biomass estimation. Especially, the estimation of canopy cover is more affected than height percentiles. Considering the entire study area, biomass estimations from both databases do not show significant differences. Simulations show that differences in biomass could be larger (more than 4%) only in particular situations, such as steep areas when the slopes are non-oriented towards the scan lines and the scan angles are larger than 15º. In chapter 3, a multi-source approach is developed, integrating available databases such as nationwide LiDAR flights, Landsat imagery and permanent field plots from SNFI, with good resultos in the generation of wall to wall forest inventories. Volume and basal area errors are similar to those obtained by other authors (using specific LiDAR flights and field plots) for the same species. Errors in the estimation of stem number are larger than literature values as a consequence of the great influence that variable-radius plots, as used in SNFI, have on this variable. In chapters 4 and 5 wall to wall plot-level methodologies to estimate aboveground biomass and carbon density in tropical forest are evaluated. The study area is located in the Poas Volcano National Park (Costa Rica) and two different situations are analyzed: i) available complete LiDAR coverage (chapter 4) and ii) a complete LiDAR coverage is not available and wall to wall estimation is carried out combining LiDAR, Landsat and ancillary data (chapter 5). In chapter 4, a general aboveground biomass plot-level LiDAR model for tropical forest (Asner & Mascaro, 2014) is validated and a specific model for the study area is fitted. Both LiDAR plot-level models are based on the top-of-canopy height (TCH) variable that is derived from the LiDAR digital canopy model. Results show that the pantropical plot-level LiDAR methodology is a reliable alternative to the development of specific models for tropical forests and thus, aboveground biomass in a new study area could be estimated by only measuring basal area (BA). Applying this methodology, the definition of precise BA field measurement procedures (e.g. location, size and shape of the field plots) is decisive to achieve reliable results in future studies. The relation between BA and TCH (Stocking Coefficient) obtained in our study area in Costa Rica varied locally. Therefore, more field work is needed for assessing Stocking Coefficient variations between different life zones and the influence of the stratification of the study areas in tropical forests on the reduction of uncertainty. In chapter 5, the combination of systematic LiDAR information sampling and full coverage Landsat imagery (and ancillary data) prove to be an effective alternative for forest inventories in tropical areas. This methodology allows estimating wall to wall vegetation height, biomass and carbon density in large areas where full LiDAR coverage and traditional field work are technically and/or economically unfeasible. Carbon density prediction using Landsat imaginery shows a slight decrease in the determination coefficient and an increase in RMSE when harshly decreasing LiDAR coverage area. Results indicate that feasible estimates of vegetation height, biomass and carbon density can be accomplished using low LiDAR coverage areas (between 5% and 20% of the total area) in tropical locations.

Brownian dynamics simulations of protein folding: Access to milliseconds time scale and beyond

Protein folding occurs on a time scale ranging from milliseconds to minutes for a majority of proteins. Computer simulation of protein folding, from a random configuration to the native structure, is nontrivial owing to the large disparity between the simulation and folding time scales. As an effort to overcome this limitation, simple models with idealized protein subdomains, e.g., the diffusion–collision model of Karplus and Weaver, have gained some popularity. We present here new results for the folding of a four-helix bundle within the framework of the diffusion–collision model. Even with such simplifying assumptions, a direct application of standard Brownian dynamics methods would consume 10,000 processor-years on current supercomputers. We circumvent this difficulty by invoking a special Brownian dynamics simulation. The method features the calculation of the mean passage time of an event from the flux overpopulation method and the sampling of events that lead to productive collisions even if their probability is extremely small (because of large free-energy barriers that separate them from the higher probability events). Using these developments, we demonstrate that a coarse-grained model of the four-helix bundle can be simulated in several days on current supercomputers. Furthermore, such simulations yield folding times that are in the range of time scales observed in experiments.

Enhanced transcription factor access to arrays of histone H3/H4 tetramer⋅DNA complexes in vitro: Implications for replication and transcription

Defined model systems consisting of physiologically spaced arrays of H3/H4 tetramer⋅5S rDNA complexes have been assembled in vitro from pure components. Analytical hydrodynamic and electrophoretic studies have revealed that the structural features of H3/H4 tetramer arrays closely resemble those of naked DNA. The reptation in agarose gels of H3/H4 tetramer arrays is essentially indistinguishable from naked DNA, the gel-free mobility of H3/H4 tetramer arrays relative to naked DNA is reduced by only 6% compared with 20% for nucleosomal arrays, and H3/H4 tetramer arrays are incapable of folding under ionic conditions where nucleosomal arrays are extensively folded. We further show that the cognate binding sites for transcription factor TFIIIA are significantly more accessible when the rDNA is complexed with H3/H4 tetramers than with histone octamers. These results suggest that the processes of DNA replication and transcription have evolved to exploit the unique structural properties of H3/H4 tetramer arrays.

A general mechanism for regulation of access to the translocon: Competition for a membrane attachment site on ribosomes

For proteins to enter the secretory pathway, the membrane attachment site (M-site) on ribosomes must bind cotranslationally to the Sec61 complex present in the endoplasmic reticulum membrane. The signal recognition particle (SRP) and its receptor (SR) are required for targeting, and the nascent polypeptide associated complex (NAC) prevents inappropriate targeting of nonsecretory nascent chains. In the absence of NAC, any ribosome, regardless of the polypeptide being synthesized, binds to the endoplasmic reticulum membrane, and even nonsecretory proteins are translocated across the endoplasmic reticulum membrane. By occupying the M-site, NAC prevents all ribosome binding unless a signal peptide and SRP are present. The mechanism by which SRP overcomes the NAC block is unknown. We show that signal peptide-bound SRP occupies the M-site and therefore keeps it free of NAC. To expose the M-site and permit ribosome binding, SR can pull SRP away from the M-site without prior release of SRP from the signal peptide.

A deoxyribonucleotidase in mitochondria: Involvement in regulation of dNTP pools and possible link to genetic disease

Three cytosolic and one plasma membrane-bound 5′-nucleotidases have been cloned and characterized. Their various substrate specificities suggest widely different functions in nucleotide metabolism. We now describe a 5′-nucleotidase in mitochondria. The enzyme, named dNT-2, dephosphorylates specifically the 5′- and 2′(3′)-phosphates of uracil and thymine deoxyribonucleotides. The cDNA of human dNT-2 codes for a 25.9-kDa polypeptide with a typical mitochondrial leader peptide, providing the structural basis for two-step processing during import into the mitochondrial matrix. The deduced amino acid sequence is 52% identical to that of a recently described cytosolic deoxyribonucleotidase (dNT-1). The two enzymes share many catalytic properties, but dNT-2 shows a narrower substrate specificity. Mitochondrial localization of dNT-2 was demonstrated by the mitochondrial fluorescence of 293 cells expressing a dNT-2-green fluorescent protein (GFP) fusion protein. 293 cells expressing fusion proteins without leader peptide or with dNT-1 showed a cytosolic fluorescence. During in vitro import into mitochondria, the preprotein lost the leader peptide. We suggest that dNT-2 protects mitochondrial DNA replication from overproduction of dTTP, in particular in resting cells. Mitochondrial toxicity of dTTP can be inferred from a severe inborn error of metabolism in which the loss of thymidine phosphorylase led to dTTP accumulation and aberrant mitochondrial DNA replication. We localized the gene for dNT-2 on chromosome 17p11.2 in the Smith–Magenis syndrome-critical region, raising the possibility that dNT-2 is involved in the etiology of this genetic disease.

WormBase: network access to the genome and biology of Caenorhabditis elegans

WormBase (http://www.wormbase.org) is a web-based resource for the Caenorhabditis elegans genome and its biology. It builds upon the existing ACeDB database of the C.elegans genome by providing data curation services, a significantly expanded range of subject areas and a user-friendly front end.

Ethics and access to teaching materials in the medical library: the case of the Pernkopf atlas*

Conflicts can occur between the principle of freedom of information treasured by librarians and ethical standards of scientific research involving the propriety of using data derived from immoral or dishonorable experimentation. A prime example of this conflict was brought to the attention of the medical and library communities in 1995 when articles claiming that the subjects of the illustrations in the classic anatomy atlas, Eduard Pernkopf's Topographische Anatomie des Menschen, were victims of the Nazi holocaust. While few have disputed the accuracy, artistic, or educational value of the Pernkopf atlas, some have argued that the use of such subjects violates standards of medical ethics involving inhuman and degrading treatment of subjects or disrespect of a human corpse. Efforts were made to remove the book from medical libraries. In this article, the history of the Pernkopf atlas and the controversy surrounding it are reviewed. The results of a survey of academic medical libraries concerning their treatment of the Pernkopf atlas are reported, and the ethical implications of these issues as they affect the responsibilities of librarians is discussed.

A study comparing centralized CD-ROM and decentralized intranet access to MEDLINE

Objective: The purpose of this study was to evaluate the efficacy of a decentralized intranet access in each medical department as opposed to centralized unique MEDLINE access in the medical library.

Molecular determinants of drug access to the receptor site for antiarrhythmic drugs in the cardiac Na+ channel.

The clinical efficacy of local anesthetic and antiarrhythmic drugs is due to their voltage- and frequency-dependent block of Na+ channels. Quaternary local anesthetic analogs such as QX-314, which are permanently charged and membrane-impermeant, effectively block cardiac Na+ channels when applied from either side of the membrane but block neuronal Na+ channels only from the intracellular side. This difference in extracellular access to QX-314 is retained when rat brain rIIA Na+ channel alpha subunits and rat heart rH1 Na+ channel alpha subunits are expressed transiently in tsA-201 cells. Amino acid residues in transmembrane segment S6 of homologous domain IV (IVS6) of Na+ channel alpha subunits have important effects on block by local anesthetic drugs. Although five amino acid residues in IVS6 differ between brain rIIA and cardiac rH1, exchange of these amino acid residues by site-directed mutagenesis showed that only conversion of Thr-1755 in rH1 to Val as in rIIA was sufficient to reduce the rate and extent of block by extracellular QX-314 and slow the escape of drug from closed channels after use-dependent block. Tetrodotoxin also reduced the rate of block by extracellular QX-314 and slowed escape of bound QX-314 via the extracellular pathway in rH1, indicating that QX-314 must move through the pore to escape. QX-314 binding was inhibited by mutation of Phe-1762 in the local anesthetic receptor site of rH1 to Ala whether the drug was applied extracellularly or intracellularly. Thus, QX-314 binds to a single site in the rH1 Na+ channel alpha subunit that contains Phe-1762, whether it is applied from the extracellular or intracellular side of the membrane. Access to that site from the extracellular side of the pore is determined by the amino acid at position 1755 in the rH1 cardiac Na+ channel.

Domino 1,3-Dipolar Cycloadditions of N-Alkyl-α-Amino Esters with Paraformaldehyde: A Direct Access to α-Hydroxymethyl α-Amino Acids

N-Alkyl-α-amino esters undergo a domino reaction, based on the iminium cation generation, with paraformaldehyde, followed by a 1,3-dipolar cycloaddition of the stabilized azometh­ine ylide with another equivalent of formaldehyde. The resulting products are oxazolidines, which can be transformed after hydrolysis into α-hydroxymethyl α-amino acid or its derivatives. The diastereoselective 1,3-dipolar cycloaddition was performed using sarcosine (–)-menthyl or (–)-8-phenylmenthyl esters affording the cyclic product with moderate enantiomeric ratio.