Regularization and Bang-bang conjugate times in optimal control

Consideramos o problema de controlo óptimo de tempo mínimo para sistemas de controlo mono-entrada e controlo afim num espaço de dimensão finita com condições inicial e final fixas, onde o controlo escalar toma valores num intervalo fechado. Quando aplicamos o método de tiro a este problema, vários obstáculos podem surgir uma vez que a função de tiro não é diferenciável quando o controlo é bang-bang. No caso bang-bang os tempos conjugados são teoricamente bem definidos para este tipo de sistemas de controlo, contudo os algoritmos computacionais directos disponíveis são de difícil aplicação. Por outro lado, no caso suave o conceito teórico e prático de tempos conjugados é bem conhecido, e ferramentas computacionais eficazes estão disponíveis. Propomos um procedimento de regularização para o qual as soluções do problema de tempo mínimo correspondente dependem de um parâmetro real positivo suficientemente pequeno e são definidas por funções suaves em relação à variável tempo, facilitando a aplicação do método de tiro simples. Provamos, sob hipóteses convenientes, a convergência forte das soluções do problema regularizado para a solução do problema inicial, quando o parâmetro real tende para zero. A determinação de tempos conjugados das trajectórias localmente óptimas do problema regularizado enquadra-se na teoria suave conhecida. Provamos, sob hipóteses adequadas, a convergência do primeiro tempo conjugado do problema regularizado para o primeiro tempo conjugado do problema inicial bang-bang, quando o parâmetro real tende para zero. Consequentemente, obtemos um algoritmo eficiente para a computação de tempos conjugados no caso bang-bang.

Mining biomedical information from scientific literature

The rapid evolution and proliferation of a world-wide computerized network, the Internet, resulted in an overwhelming and constantly growing amount of publicly available data and information, a fact that was also verified in biomedicine. However, the lack of structure of textual data inhibits its direct processing by computational solutions. Information extraction is the task of text mining that intends to automatically collect information from unstructured text data sources. The goal of the work described in this thesis was to build innovative solutions for biomedical information extraction from scientific literature, through the development of simple software artifacts for developers and biocurators, delivering more accurate, usable and faster results. We started by tackling named entity recognition - a crucial initial task - with the development of Gimli, a machine-learning-based solution that follows an incremental approach to optimize extracted linguistic characteristics for each concept type. Afterwards, Totum was built to harmonize concept names provided by heterogeneous systems, delivering a robust solution with improved performance results. Such approach takes advantage of heterogenous corpora to deliver cross-corpus harmonization that is not constrained to specific characteristics. Since previous solutions do not provide links to knowledge bases, Neji was built to streamline the development of complex and custom solutions for biomedical concept name recognition and normalization. This was achieved through a modular and flexible framework focused on speed and performance, integrating a large amount of processing modules optimized for the biomedical domain. To offer on-demand heterogenous biomedical concept identification, we developed BeCAS, a web application, service and widget. We also tackled relation mining by developing TrigNER, a machine-learning-based solution for biomedical event trigger recognition, which applies an automatic algorithm to obtain the best linguistic features and model parameters for each event type. Finally, in order to assist biocurators, Egas was developed to support rapid, interactive and real-time collaborative curation of biomedical documents, through manual and automatic in-line annotation of concepts and relations. Overall, the research work presented in this thesis contributed to a more accurate update of current biomedical knowledge bases, towards improved hypothesis generation and knowledge discovery.

Infection mechanism of Diplodia corticola

Diplodia corticola is regarded as the most virulent fungus involved in cork oak decline, being able to infect not only Quercus species (mainly Q. suber and Q. ilex), but also grapevines (Vitis vinifera) and eucalypts (Eucalyptus sp.). This endophytic fungus is also a pathogen whose virulence usually manifests with the onset of plant stress. Considering that the infection normally culminates in host death, there is a growing ecologic and socio-economic concern about D. corticola propagation. The molecular mechanisms of infection are hitherto largely unknown. Accordingly, the aim of this study was to unveil potential virulence effectors implicated in D. corticola infection. This knowledge is fundamental to outline the molecular framework that permits the fungal invasion and proliferation in plant hosts, causing disease. Since the effectors deployed are mostly proteins, we adopted a proteomic approach. We performed in planta pathogenicity tests to select two D. corticola strains with distinct virulence degrees for our studies. Like other filamentous fungi D. corticola secretes protein at low concentrations in vitro in the presence of high levels of polysaccharides, two characteristics that hamper the fungal secretome analysis. Therefore, we first compared several methods of extracellular protein extraction to assess their performance and compatibility with 1D and 2D electrophoretic separation. TCA-Acetone and TCA-phenol protein precipitation were the most efficient methods and the former was adopted for further studies. The proteins were extracted and separated by 2D-PAGE, proteins were digested with trypsin and the resulting peptides were further analysed by MS/MS. Their identification was performed by de novo sequencing and/or MASCOT search. We were able to identify 80 extracellular and 162 intracellular proteins, a milestone for the Botryosphaeriaceae family that contains only one member with the proteome characterized. We also performed an extensive comparative 2D gel analysis to highlight the differentially expressed proteins during the host mimicry. Moreover, we compared the protein profiles of the two strains with different degrees of virulence. In short, we characterized for the first time the secretome and proteome of D. corticola. The obtained results contribute to the elucidation of some aspects of the biology of the fungus. The avirulent strain contains an assortment of proteins that facilitate the adaptation to diverse substrates and the identified proteins suggest that the fungus degrades the host tissues through Fenton reactions. On the other hand, the virulent strain seems to have adapted its secretome to the host characteristics. Furthermore, the results indicate that this strain metabolizes aminobutyric acid, a molecule that might be the triggering factor of the transition from a latent to a pathogenic state. Lastly, the secretome includes potential pathogenicity effectors, such as deuterolysin (peptidase M35) and cerato-platanin, proteins that might play an active role in the phytopathogenic lifestyle of the fungus. Overall, our results suggest that D. corticola has a hemibiotrophic lifestyle, switching from a biotrophic to a necrotrophic interaction after plant physiologic disturbances.This understanding is essential for further development of effective plant protection measures.