Thèse numérisée par la Direction des bibliothèques de l'Université de Montréal.

Fundulus is a diverse and widespread genus of small teleost fish of North America. Due to its high tolerance for physiochemical variation (e.g. temperature, oxygen, salinity), Fundulus is a model organism to study physiological and molecular adaptations to environmental stress. The thesis focuses on patterns of protein expression in Fundulus heteroclitus and F. grandis.The patterns of protein expression were investigated using traditional methods of enzyme activity measurements and recent proteomic approaches. The findings of the study can be used to guide future studies on the proteomic responses of vertebrates to environmental stress. Chapter 2 focuses on measurement of the temporal effects of oxygen treatments on the maximal specific activities of nine glycolytic enzymes in liver and skeletal muscle during chronic exposure (28d) of Fundulus heteroclitus. The fish was exposed to four different oxygen treatments: hyperoxia, normoxia, moderate hypoxia, and severe hypoxia. The time course of changes in maximal glycolytic enzyme specific activities was assessed at 0, 8, 14 and 28 d. The results demonstrate that chronic hypoxia alters the capacity for carbohydrate metabolism in F. heteroclitus, with the important observation that the responses are both tissue- and enzyme-specific. Chapter 3 studies the effect of tissue storage on protein profile of tissues of F. grandis. The technique of one dimensional gel electrophoresis (1D-SDS-PAGE) was used to assess the effects of tissue sampling, flash frozen in liquid nitrogen versus immersion of fresh tissue in RNA later, for five tissues, liver, skeletal muscle, brain, gill, and heart, followed by LC-MS/MS to identify protein bands that were differentially stabilized in gill and liver. The study shows that, in F. grandis, the preferred method of preservation was tissue specific. xi Chapter 4 focuses on the use of advanced 2DE-MS/MS to characterize the proteome of multiple tissues in F. grandis. Database searching resulted in the identification of 253 non-redundant proteins in five tissues: liver, muscle, brain, gill, and heart. Identifications include enzymes of energy metabolism, heat shock proteins, and structural proteins. The protein identification rate was approximately 50 % of the protein spots analyzed. This identification rate for a species without a sequenced genome demonstrates the utility of F. grandis as a model organism for environmental proteomic studies in vertebrates.

Η θερμοκρασία ανάπτυξης αποτελεί παράγοντα μεγάλης σημασίας στην οντογένεση των ιχθύων, αφού ως ποικιλόθερμοι οργανισμοί είναι συνεχώς εκτεθειμένοι στις μεταβολές του περιβάλλοντός τους. Έχει παρατηρηθεί πως η θερμοκρασία ανάπτυξης δύναται να προκαλέσει μετατόπιση του χρονοδιαγράμματος των οντογενετικών γεγονότων και πλαστικότητα σε μορφολογικούς και φυσιολογικούς χαρακτήρες (πχ στο μυοσκελετικό και το καρδιαγγειακό σύστημα). Ωστόσο, μέχρι σήμερα δεν έχει μελετηθεί η επίδραση της θερμοκρασίας ανάπτυξης στο πρότυπο της γονιδιακής έκφρασης του zebrafish και σκοπός της παρούσας εργασίας είναι η μελέτη του ολικού μεταγραφικού προτύπου νυμφών zebrafish. Για το λόγο αυτό σχεδιάστηκαν δύο πειράματα, όπου τρεις θερμοκρασιακές συνθήκες ανάπτυξης (22, 28 και 32oC) εφαρμόστηκαν στην πρώιμη οντογενετική περίοδο: για το διάστημα 0-20 dpf* (1ο πείραμα) και 10-20 dpf (2ο πείραμα). Πραγματοποιήθηκε απομόνωση ολικού RNA από τα άτομα ηλικίας 20 dpf όλων των πληθυσμών και από τα άτομα ηλικίας 10 dpf του πληθυσμού των 28oC του 2ου πειράματος και ακολούθησε υβριδοποίηση σε ολιγονουκλεοτιδικές μικροσυστοιχίες Affymetrix, με 15.509 αντιπροσωπευτικές αλληλουχίες γονιδίων (probe sets). Τα 21 μεταγραφικά προφίλ επεξεργάσθηκαν με τα εξειδικευμένα προγράμματα ανάλυσης μικροσυστοιχιών, dChip και MeV (v.4.5.1). Η κανονικοποίηση και το φιλτράρισμα των δεδομένων των μικροσυστοιχιών απέδωσε μεταγραφικά πρότυπα με 9.488 probe sets. Με τεχνικές πολυπαραμετρικής στατιστικής ανάλυσης (HCL και PCA) πραγματοποιήθηκαν οι συγκρίσεις των μεταγραφικών προτύπων μεταξύ των πειραματικών πληθυσμών για κάθε θερμοκρασία ανάπτυξης και οντογενετικό στάδιο. Οι HCL και PCA αναλύσεις έδειξαν i) σαφή διαχωρισμό των μεταγραφικών προτύπων μεταξύ των δύο πειραμάτων, ii) σαφή διαχωρισμό των προτύπων των 28oC και 32oC ως προς αυτά των 22oC και στα δύο πειράματα, και iii) σαφή διαχωρισμό των προτύπων των 28oC διαφορετικού οντογενετικού σταδίου (ηλικίας 10 dpf vs 20 dpf). Θα αναμέναμε τα πρότυπα έκφρασης των 28oC να παρουσιάζουν παρόμοιο πρότυπο, κάτι που δεν παρατηρείται. Αυτό οφείλεται στο πειραματικό σφάλμα που υπεισέρχεται από το διαφορετικό χρόνο πραγματοποίησης των δύο πειραμάτων και τις ρυθμίσεις κατά την υβριδοποίηση. Έτσι πραγματοποιήθηκε η κανονικοποίηση των “28”, που απαλείφει το πειραματικό σφάλμα. Οι HCL και PCA αναλύσεις έδειξαν i) σαφή διαχωρισμό των μεταγραφικών προτύπων των 28oC και 32oC ως προς αυτά των 22oC ως αποτέλεσμα της επίδρασης της θερμοκρασίας ανάπτυξης, ii) σαφή διαχωρισμό των προτύπων των 22oC των δύο πειραμάτων ως αποτέλεσμα της επίδρασης της περιόδου εφαρμογής και διάρκειας της θερμοκρασιακής αγωγής και iii) σαφή διαχωρισμό των πρότυπων των 28oC (ηλικίας 10 dpf vs 20 dpf) ως αποτέλεσμα της επίδρασης του οντογενετικού σταδίου. Ακολούθως, πραγματοποιήθηκε ανάλυση σημαντικότητας (SAM) και λειτουργική γονιδιωματική ανάλυση (λογισμικό DAVID) των στατιστικώς σημαντικών γονιδίων που διαφοροποιούν τα πρότυπα. Η ανάλυση των γονιδίων, ως προς την ιστοειδική έκφραση ανέδειξε γονίδια που σχετίζονται με την ανάπτυξη του ματιού, των θωρακικών πτερυγίων και του εγκεφάλου στους πληθυσμούς των 22oC έναντι των 28oC και 32oC. Η παρούσα εργασία αποδεικνύει την επίδραση της θερμοκρασίας ανάπτυξης και της διάρκειας της θερμοκρασιακής αγωγής, καθώς επάγει μηχανισμούς πλαστικότητας στο επίπεδο της γονιδιακής έκφρασης. Τέλος, υποδεικνύεται πως η πρώιμη οντογενετική περίοδος τείνει να είναι περισσότερο θερμοευαίσθητη, καθώς παρατηρείται εντονότερη επίδραση της θερμοκρασίας (στην περίπτωση των 22οC)στην ανάπτυξη του ατόμου. *dpf: days post fertilization / Developmental temperature plays a principal role in the ontogeny of fish. It is known that developmental temperature may shift the initiation time of the ontogenetic stages and induce plasticity in morphological and physiological characters e.g. the musculoskeletal and the cardiovascular system. However, its effect on the gene expression pattern has not previously been attempted for zebrafish. In the present study, zebrafish Affymetrix microarrays of 15,509 probe sets were used to map the transcriptome profile of: a) 20 dpf* old zebrafish larvae at three developmental temperatures, i.e. 22oC, 28oC and 32oC (1st experiment) and b) 20 dpf old zebrafish larvae, which were all grown at 28oC for the first 10 days and subsequently divided into three groups, which were grown at 22oC, 28oC and 32oC, respectively; the profile of 10 dpf old larvae was also measured (2nd experiment). We have isolated total RNA from the above populations and then, hybridization of RNA samples has been done on oligonucleotide Affymetrix microarrays of 15,509 probe sets. All 21 profiles were normalized and filtered (dChip software), and multivariate statistical analysis techniques were used on the normalized 9,488 probe set expression profiles (TM4 MeV software). Hierarchical Clustering (HCL) and Principal Component Analysis (PCA) on expression profiles indicated: a) clear separation of the two experiments based on their transcriptomic patterns, b) clustering of the 28oC and 32oC profiles of the 20 dpf old larvae separately from those at 22oC in both experiments and c) clear separation of the 28oC profiles based on the developmental stage. We would expect expression profiles of 28oC to be clustered together, though this was not observed because of experimental parameters during the hybridization, as the two experiments were carried out independently on different dates. So the normalization of “28” profiles took place, in order to eliminate the experimental noise. HCL and PCA, then, indicated: a) clustering of the 28oC and 32oC profiles of the 20 dpf old larvae separately from those at 22oC, as the effect of developmental temperature, b) clear separation of 22oC profiles of the two experiments, based on the effect of the period and duration of thermal conditions and c) clear separation of the 28oC profiles based on the developmental stage. Then, Significant Analysis of Microarrays (SAM) and Functional Genomic Classification Analysis (DAVID software) of statistically significant genes was carried out. Analysis of genes based on tissue-specific expression indicated characteristic genes for the development of the eye, pectoral fins and brain in 22oC profiles versus 28oC and 32oC profiles. The present study has proved that thermal effect is determinative among the early ontogenetic stage, especially in the case of longer cold thermal period, and developmental temperature may induce plastic response of gene expression, that could affect the fate of fish. *dpf: days post fertilization

Made available in DSpace on 2016-06-02T19:03:58Z (GMT). No. of bitstreams: 1 6030.pdf: 2345815 bytes, checksum: 385c6d8c1bda1d4afe540c01668338fa (MD5) Previous issue date: 2013-10-08 / Nowadays, the amount of biological data available by universities, hospitals and research centers has increased exponentially due the use of bioinformatics, with the development of methods and advanced computational tools, and high-throughput techniques. Due to this significant increase in the amount of available data, new strategies for capture, storage and analysis of data are necessary. In this scenario, a new research area is developing, called biocuration. The biocuration is becoming a fundamental part in the biological and biomedical research, and the main function is related with the structuration and organization of the biological information, making it readable and accessible to mens and computers. Seeking to support a fast and reliable understanding of new domains, different initiatives are being proposed, and the Gene Ontology (GO) is one of the main examples. The GO is one the main initiatives in bioinformatics, whose main goal is to standardize the representation of genes and their products, providing interconnections between species and databases. Thus, the main objective of this research is to propose a computational architecture that uses principles of never-ending learning to help biocurators in new GO classifications. Nowadays, this classification task is totally manual. The proposed architecture uses semi-supervised learning combining different classifiers used in the classification of new GO samples. In addition, this research also aims to build high-level knowledge in the form of simple IF-THEN rules and decision trees. The generated knowledge can be used by the GO biocurators in the search for important patterns present in the biological data, revealing concise and relevant information about the application domain. / Nos dias atuais, a quantidade de dados biológicos disponibilizados por universidades, hospitais e centros de pesquisa tem aumentado de forma exponencial, devido ao emprego da bio-informática, através do desenvolvimento de métodos e técnicas computacionais avançados, e de técnicas de high-throughput. Devido a esse significativo aumento na quantidade de dados disponibilizados, gerou-se a necessidade da criação de novas estratégias para captura, armazenamento e principalmente analise desses dados. Devido a esse cenário, um novo campo de trabalho e pesquisa vem surgindo, chamado biocuragem. A biocuragem está se tornando parte fundamental na pesquisa biomédica e biológica, e tem por principal função estruturar e organizar a informação biológica, tornando-a legível e acessível a homens e computadores. Buscando prover um rápido e confiável entendimento de novos domínios, diferentes iniciativas estão sendo propostas, tendo no Gene Ontology (GO) um dos seus principais exemplos. O GO se destaca mundialmente sendo uma das principais iniciativas em bioinformática, cuja principal meta e padronizar a representação dos genes e seus produtos, provendo interconexões entre espécies e bancos de dados. Dessa forma, objetiva-se com essa pesquisa propor uma arquitetura computacional que utiliza princípios de aprendizado de maquina sem-fim para auxiliar biocuradores do GO na tarefa de classificação de novos termos, tarefa essa, totalmente manual. A arquitetura proposta utiliza aprendizado semi-supervisionado combinando diferentes classificadores na rotulação de novas instâncias do GO. Além disso, essa pesquisa também tem por objetivo a construção de conhecimento de alto-nível na forma de simples regras SE-ENTÃO e árvores de decisão. Esse conhecimento gerado pode ser utilizado pelos biocuradores do GO na busca por padrões importantes presentes nos dados biológicos, revelando informações concisas e relevantes sobre o domínio da aplicação.

Submitted by Cláudia Bueno (claudiamoura18@gmail.com) on 2016-06-09T17:05:06Z No. of bitstreams: 2 Tese - Letícia Mendes de Castro - 2016.pdf: 2261024 bytes, checksum: 4cfaccebd66c7e3ea4eda9b31d53d16d (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Approved for entry into archive by Luciana Ferreira (lucgeral@gmail.com) on 2016-06-10T11:19:37Z (GMT) No. of bitstreams: 2 Tese - Letícia Mendes de Castro - 2016.pdf: 2261024 bytes, checksum: 4cfaccebd66c7e3ea4eda9b31d53d16d (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Made available in DSpace on 2016-06-10T11:19:37Z (GMT). No. of bitstreams: 2 Tese - Letícia Mendes de Castro - 2016.pdf: 2261024 bytes, checksum: 4cfaccebd66c7e3ea4eda9b31d53d16d (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Previous issue date: 2016-03-04 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Brazil has one of the largest commercial cattle herds worldwide, but its meat quality is highly variable. The national herd is largely composed of Bos indicus breeds, which in general have less tender meat than Bos taurus cattle, decreasing the product value. This study was carried out to identify relevant regions and biological pathways associated with meat tenderness in Polled Nellore cattle. It was also aimed to evaluate the effect of different quality control protocols in GWAS for meat tenderness in Polled Nellore cattle. The data consisted of Warner-Bratzler shear force (WBSF) values of Longissimus muscle after 7 days of ageing, from 427 Polled Nellore animals. The animals were genotyped using either the Illumina BovineHD Beadchip (777k) or the GGP-Indicus Chip (77k). SNPs were excluded when Call Rate < 90%, then the imputation from the GGP to the HD Chip was performed using the FImput’s software. To study the different quality control protocols and their influence in GWAS, 590,915 markers were used. The effect of different QCs were verified using 16 protocols with three thresholds for MAF (MAF < 0.01; < 0.05 and < 0.10) and HWP (p < 0.01; < 0.0001 and < 0.00001) and their possible combinations. GWASs were performed using the PD3/EMMAx method with the remaining markers of each QC. For GWAS performed for pathway analysis, 369,007 markers were used after SNPs were excluded when Call rate < 90%, HWP p < 0.01 and MAF < 0.01. Group of slaughter and sex were included as fixed effects. Significant markers (p < 0.0001) were found in all analysis, in which the chromosomes with more significant SNPs of the different QCs were 3, 17, 20, 21, 25 and 27, and in the pathway study were located on chromosomes 3, 13, 17, 20, 21 and 25 explaining great proportion of variation, indicating possible QTLs associated with meat tenderness in those genomic regions. The analyses of different QCs showed that there is an effect of quality control over GWAS, and the filter for MAF influenced the results more broadly. A pathway enrichment analysis based on SNPs from GWAS was performed using FatiGO’s procedure. 22,365 annotated genes, including 1,010 significant genes were used. Thus, 22 GO terms and two IP entries were deemed enriched. Several of these functional categories, such as protein tyrosine and serine/threonine kinase activity, calcium ion binding and growth factors can be related to WBSF in Polled Nellore cattle. These results help to elucidate the metabolic pathways related to this trait, which is of extreme economic and social importance to Brazil as Nellore is the dominant beef cattle breed in the country. / O Brasil tem um dos maiores rebanhos bovinos comerciais do mundo, mas a qualidade da carne é altamente variável. O rebanho nacional é em grande parte composto de raças Bos indicus, que em geral têm carne menos macia do que o gado Bos taurus, diminuindo o valor do produto. Objetivou-se nesse estudo identificar regiões genômicas e vias biológicas relevantes associadas com a maciez da carne em bovinos da raça Nelore Mocho. Além disso, objetivou-se também avaliar diferentes protocolos de controle de qualidade dos SNPs e as possíveis influências nos resultados de GWAS. Os dados consistiram em valores de WBSF do músculo Longissimus dorsi, após maturação de sete dias, de 427 animais Nelore Mocho. Os animais foram genotipados em marcadores SNP Illumina BovineHD Beadchip (777k) ou Chip GGP-Indicus (77k). Todos os SNPs passaram por um Call Rate de 90% para posterior imputação utilizando o software FImput. Para averiguar os diferentes protocolos de qualidade e suas influências no GWAS foram utilizados 590.915 marcadores. Os efeitos dos diferentes QCs foram verificados utilizando 16 protocolos com três limiares para MAF (MAF < 0,01;< 0,05 e < 0,10) e HWP (p < 0,01; < 0,0001 e < 0,00001) e suas possíveis combinações. Os GWASs foram realizados utilizando método P3D/EMMAx com os marcadores restantes de cada QC. No GWAS realizado para posterior análise das vias utilizou-se 369.007 marcadores após a exclusão de SNPs baseada nos filtros Call Rate < 90%, HWP p < 0,01 e MAF < 0,01. Grupo de abate e sexo foram incluídos no modelo como efeitos fixos. Marcadores significativos (p < 0,0001) foram localizados em todas as análises, e os cromossomos com maior quantidade de SNPs significativos dos diferentes QCs foram 3, 17, 20, 21, 25 e 27. No estudo de vias foram localizados SNPs significativos nos cromossomos 3, 13, 17, 20, 21 e 25, que explicaram maior proporção da variação, indicando que existem QTLs associados à maciez da carne nessas regiões do genoma. As análises dos diferentes QCs evidenciaram efeito do controle de qualidade dos SNPs sobre o GWAS e o filtro para MAF influenciou de maneira mais ampla os resultados. Foi realizada uma análise de enriquecimento de vias baseando-se nos SNPs do GWAS, utilizando o procedimento FatiGO. Apenas os genes com no mínimo um SNP significativo (p < 0,01) foram considerados. Foram utilizados 22.365 genes anotados, incluindo 1.010 genes significativos. Um total de 22 termos GO e duas entradas IP foram consideradas enriquecidas com genes significamente associados com a maciez da carne. Várias dessas categorias funcionais como atividade da proteína tirosina quinase e serina/treonina quinase, ligantes ao íon cálcio e fatores de crescimento, podem estar relacionadas com WBFS em bovinos da raça Nelore Mocho. Estes resultados ajudam elucidar as vias relacionadas com essa característica de extrema importância econômica para o Brasil, já que o Nelore é a raça de gado de corte dominante no país.

No description available.

Abiotic stress is a major constraint for crop productivity worldwide. To better understand the common biological mechanisms of abiotic stress responses in plants, we performed meta-analysis of 652 samples of RNA sequencing (RNA-Seq) data from 43 published abiotic stress experiments in Arabidopsis thaliana. These samples were categorized into eight different abiotic stresses including drought, heat, cold, salt, light and wounding. We developed a multi-step computational pipeline, which performs data downloading, preprocessing, read mapping, read counting and differential expression analyses for RNA-Seq data. We found that 5729 and 5062 genes are induced or repressed by only one type of abiotic stresses. There are only 18 and 12 genes that are induced or repressed by all stresses. The commonly induced genes are related to gene expression regulation by stress hormone abscisic acid. The commonly repressed genes are related to reduced growth and chloroplast activities. We compared stress responsive genes between any two types of stresses and found that heat and cold regulate similar set of genes. We also found that high light affects different set of genes than blue light and red light. Interestingly, ABA regulated genes are different from those regulated by other stresses. Finally, we found that membrane related genes are repressed by ABA, heat, cold and wounding but are up regulated by blue light and red light. The results from this work will be used to further characterize the gene regulatory networks underlying stress responsive genes in plants. / Master of Science

Ein der Biologie zugrunde liegender Prozess ist die Fortpflanzung. Einzeller wachsen dazu heran und teilen sich. Grundlage hierfür sind ausreichend Nahrung und Ressourcen, um die eigene Masse und alle Zellbestandteile, insbesondere die DNS, zu verdoppeln. Fehler bei der Wachstumsregulation oder der DNS-Verdopplung können schwerwiegende Folgen haben und stehen beim Menschen im Zusammenhang z.B. mit Krebs. In dieser Arbeit werden mathematische Modelle für die Mechanismen zur Wachstumsregulierung und DNS-Verdopplung in der Bäckerhefe, Saccharomyces cerevisiae, vorgestellt. Modellierung kann entscheident zum Verstehen von komplexen, dynamischen Systemen beitragen. Wir haben ein Modell für Einzellerwachstum entwickelt und leiten das Wachstumsverhalten von Zellkulturen von diesem Modell, mittels einer hierfür programmierten Software, ab. Außerdem haben wir ein Model für die Verdopplung der DNS entwickelt, um Auswirkungen verschiedener Aktivierungsmuster auf die Replikation zu testen. Zusätzlich wurde die Verlängerung entstehender DNS Stränge, Elongation, mit einem detaillierten, stochastischen Modell untersucht. Wir haben unsere Ergebnisse zur DNS-Verdopplung mit einer abschließenden Untersuchung ergänzt, die funktionelle Beziehungen von Genen aufzeigt, welche sich in der Nähe von Aktivierungsstellen der Verdopplung befinden. Folgende Einsichten in die komplexe Koordination von Wachstum und Teilung wurden gewonnen: (i) Wachstumskontrolle ist eine inhärente Eigenschaft von Hefezellpopulationen, welche weder Signale noch Messmechanismen benötigt, (ii) DNS Verdopplung ist robuster in kleinen Chromosomen mit hoher Dichte an Aktivierungsstellen, (iii) Elongation ist weitgehend uniform, weicht aber an genau definierten Stellen signifikant ab und (iv) katabole Gene häufen sich nahe der frühen Aktivierungsstellen und anabole Gene nahe der späten. Unsere Ergebnisse tragen zum Verständniss von zellulären Mechanismen zur Wachstumskontrolle und DNS-Verdopplung bei. / One of the most fundamental processes in biology is reproduction. To achieve this, single cellular organisms grow, proliferate and divide. The prerequisite for this is acquiring sufficient resources to double size and cellular components, most importantly the DNA. Defects in either sufficient gain in size or chromosomal doubling can be severe for the organism and have been related to diseases in humans, such as cancer. Therefore, the cell has developed sophisticated regulatory mechanisms to control the orderly fashion of growth and duplication. We have developed mathematical models to study systemic properties of size control and DNA replication in the premier eukaryotic model organism Saccharomyces cerevisiae. Modeling can help understanding the complex nature of dynamic systems. We provide a single cell model to explore size control. We deduced population behavior from the single cell model through multi-cell simulations using a tailor-made software. Also, we implemented an algorithm that simulates the DNA replication process to test the impact of different replication activation patterns. Additionally, elongation dynamics were assessed with a fine-grained stochastic model for the replication machinery motion. We complemented our analysis of DNA replication by studying the functional association of genes and replication origins. Our systems-level analysis reveals novel insights into the coordination of growth and division, namely that (i) size regulation is an intrinsic property of yeast cell populations and not due to signaling or size sensing, (ii) DNA replication is more robust in small chromosomes with high origin density, (iii) the elongation process is strongly biased at distinct locations in the genome and (iv) catabolic genes are over-represented near early origins and anabolic genes near late origins. Our results contribute to explaining mechanisms of size control and DNA replication.

Background: Most of our biomedical knowledge is only accessible through texts. The biomedical literature grows exponentially and PubMed comprises over 18.000.000 literature abstracts. Recently much effort has been put into the creation of biomedical ontologies which capture biomedical facts. The exploitation of ontologies to explore the scientific literature is a new area of research. Motivation: When people search, they have questions in mind. Answering questions in a domain requires the knowledge of the terminology of that domain. Classical search engines do not provide background knowledge for the presentation of search results. Ontology annotated structured databases allow for data-mining. The hypothesis is that ontology annotated literature databases allow for text-mining. The central problem is to associate scientific publications with ontological concepts. This is a prerequisite for ontology-based literature search. The question then is how to answer biomedical questions using ontologies and a literature corpus. Finally the task is to automate bibliometric analyses on an corpus of scientific publications. Approach: Recent joint efforts on automatically extracting information from free text showed that the applied methods are complementary. The idea is to employ the rich terminological and relational information stored in biomedical ontologies to markup biomedical text documents. Based on established semantic links between documents and ontology concepts the goal is to answer biomedical question on a corpus of documents. The entirely annotated literature corpus allows for the first time to automatically generate bibliometric analyses for ontological concepts, authors and institutions. Results: This work includes a novel annotation framework for free texts with ontological concepts. The framework allows to generate recognition patterns rules from the terminological and relational information in an ontology. Maximum entropy models can be trained to distinguish the meaning of ambiguous concept labels. The framework was used to develop a annotation pipeline for PubMed abstracts with 27,863 Gene Ontology concepts. The evaluation of the recognition performance yielded a precision of 79.9% and a recall of 72.7% improving the previously used algorithm by 25,7% f-measure. The evaluation was done on a manually created (by the original authors) curation corpus of 689 PubMed abstracts with 18,356 curations of concepts. Methods to reason over large amounts of documents with ontologies were developed. The ability to answer questions with the online system was shown on a set of biomedical question of the TREC Genomics Track 2006 benchmark. This work includes the first ontology-based, large scale, online available, up-to-date bibliometric analysis for topics in molecular biology represented by GO concepts. The automatic bibliometric analysis is in line with existing, but often out-dated, manual analyses. Outlook: A number of promising continuations starting from this work have been spun off. A freely available online search engine has a growing user community. A spin-off company was funded by the High-Tech Gründerfonds which commercializes the new ontology-based search paradigm. Several off-springs of GoPubMed including GoWeb (general web search), Go3R (search in replacement, reduction, refinement methods for animal experiments), GoGene (search in gene/protein databases) are developed.

With the ever increase in biomedical literature, text-mining has emerged as an important technology to support bio-curation and search. Word sense disambiguation (WSD), the correct identification of terms in text in the light of ambiguity, is an important problem in text-mining. Since the late 1940s many approaches based on supervised (decision trees, naive Bayes, neural networks, support vector machines) and unsupervised machine learning (context-clustering, word-clustering, co-occurrence graphs) have been developed. Knowledge-based methods that make use of the WordNet computational lexicon have also been developed. But only few make use of ontologies, i.e. hierarchical controlled vocabularies, to solve the problem and none exploit inference over ontologies and the use of metadata from publications. This thesis addresses the WSD problem in biomedical ontologies by suggesting different approaches for word sense disambiguation that use ontologies and metadata. The "Closest Sense" method assumes that the ontology defines multiple senses of the term; it computes the shortest path of co-occurring terms in the document to one of these senses. The "Term Cooc" method defines a log-odds ratio for co-occurring terms including inferred co-occurrences. The "MetaData" approach trains a classifier on metadata; it does not require any ontology, but requires training data, which the other methods do not. These approaches are compared to each other when applied to a manually curated training corpus of 2600 documents for seven ambiguous terms from the Gene Ontology and MeSH. All approaches over all conditions achieve 80% success rate on average. The MetaData approach performs best with 96%, when trained on high-quality data. Its performance deteriorates as quality of the training data decreases. The Term Cooc approach performs better on Gene Ontology (92% success) than on MeSH (73% success) as MeSH is not a strict is-a/part-of, but rather a loose is-related-to hierarchy. The Closest Sense approach achieves on average 80% success rate. Furthermore, the thesis showcases applications ranging from ontology design to semantic search where WSD is important.