RNA-Seq is an increasingly popular approach to transcriptome profiling that uses the capabilities of next generation sequencing technologies and provides better measurement of levels of transcripts and their isoforms. In this thesis, we apply RNA-Seq protocol and transcriptome quantification to estimate gene expression and pathway activity levels. We present a novel method, called IsoDE, for differential gene expression analysis based on bootstrapping. In the first version of IsoDE, we compared the tool against four existing methods: Fisher's exact test, GFOLD, edgeR and Cuffdiff on RNA-Seq datasets generated using three different sequencing technologies, both with and without replicates. We also introduce the second version of IsoDE which runs 10 times faster than the first implementation due to some in-memory processing applied to the underlying gene expression frequencies estimation tool and we also perform more optimization on the analysis. The second part of this thesis presents a set of tools to differentially analyze metabolic pathways from RNA-Seq data. Metabolic pathways are series of chemical reactions occurring within a cell. We focus on two main problems in metabolic pathways differential analysis, namely, differential analysis of their inferred activity level and of their estimated abundance. We validate our approaches through differential expression analysis at the transcripts and genes levels and also through real-time quantitative PCR experiments. In part Four, we present the different packages created or updated in the course of this study. We conclude with our future work plans for further improving IsoDE 2.0.

Estudos de expressão gênica geralmente envolvem a realização de processos de análise integrados para a obtenção de respostas biológicas de interesse. A realização destes processos frequentemente requer o uso combinado de uma série de ferramentas de software. No entanto, o processo de integração manual de ferramentas pode ser demorado e propenso a erros devido ao crescente número de ferramentas e formatos de dados disponíveis no domínio. De modo a automatizar o processo de integração, algumas abordagens têm sido propostas tanto para a adaptação das ferramentas de análise existentes como serviços web semânticos, quanto para o desenvolvimento de ambientes de suporte à integração (composição) de serviços web semânticos. Embora estas abordagens representem avanços, nenhuma solução adequada para o desenvolvimento e composição de serviços foi especificamente definida para o domínio de genômica funcional. Neste contexto, o principal objetivo deste projeto foi investigar uma solução completa para o desenvolvimento e composição de serviços web semânticos para a análise de expressão gênica. Como parte da solução proposta, definimos uma metodologia integrada para a implementação de serviços web semânticos criados a partir de ferramentas de software existentes e para a anotação semântica destes serviços. Nossa metodologia fornece diretrizes concretas para o desenvolvimento sistemático de serviços, considerando também os principais aspectos técnicos associados ao processo de desenvolvimento. Esta metodologia foi aplicada a um conjunto representativo de serviços que fornecem suporte às principais atividades de análise realizadas em diferentes tipos de dados de expressão gênica. De forma complementar, definimos uma solução completa para a composição semântica de serviços no domínio de análise de expressão gênica. A solução proposta foi implementada em uma plataforma de suporte semi-automático à composição de serviços web semânticos, chamada SemanticSCo. Esta plataforma fornece suporte flexível a todas as atividades envolvidas no processo de composição de serviços, incluindo a criação, publicação, requisição, descoberta, seleção, composição e execução de serviços. Além disto, a plataforma SemanticSCo foi projetada para prover suporte adequado a diferentes tipos de usuários, incluindo biologistas e bioinformatas. Neste sentido, a plataforma fornece aos usuários um alto nível de abstração para a definição de seus processos de análise, permitindo que os mesmos se concentrem mais nas questões de pesquisa biológicas do que nos aspectos subjacentes do processo de composição. Adicionalmente, a plataforma SemanticSCo suporta a definição e incorporação não apenas de serviços simples, definidos em termos de uma única operação, mas também de serviços complexos, definidos em termos de um conjunto de condições que restringem a ordem de invocação de suas operações. Finalmente, de modo a avaliar a plataforma de suporte desenvolvida, definimos diferentes cenários de composição para a análise (integrada) de dados de expressão gênica. O uso da plataforma SemanticSCo facilitou a definição destes cenários, permitindo assim a reprodução dos resultados obtidos a partir de diferentes estudos de expressão gênica previamente documentados na literatura / Gene expression studies usually involve the creation of integrated analysis processes for obtaining responses for a biological question. The creation of such processes often require the combined use of a number of software tools. However, the manual integration of tools can be cumbersome and error prone due to the increasing number of tools and data formats available in the domain. In order to automate the integration process, some approaches have been proposed for the adaptation of existing analysis tools as semantic web services as well as for the development of software environments to support the integration (composition) of semantic web services. Although these approaches present advances, to the best of our knowledge, no suitable solution has been proposed for the development and composition of web services in the functional genomics domain. In this context, this project aimed at investigating a complete solution for the development and composition of semantic web services to support gene expression analysis. As part of the proposed solution, we have defined an integrated methodology for the implementation of semantic web services created from existing software tools and the semantic annotation of such services. Our methodology provides concrete guidelines for the systematic development of services, also taking into account the main technical aspects associated with the development process. This methodology has been applied in the development of a representative set of services that support the main analysis activities performed on different types of gene expression data. Complementary to our methodology, we have defined a complete solution for the semantic composition of web services in the gene expression analysis domain. The proposed solution has been implemented in a software platform to support the semi-automatic composition of semantic web services, named SemanticSCo. This platform provides flexible support to all activities involved in the service composition process including service creation, publication, request, discovery, selection, composition and execution. Additionally, the SemanticSCo platform has been designed to support different types of users, including biologists and bioinformaticians. In this sense, the platform provides users with a high level of abstraction in the definition of their analysis processes, thus allowing them to focus more on biological research issues rather than on underlying details of the composition process. In addition, the SemanticSCo platform supports not only the definition and incorporation of (simple) services defined in terms of a single operation, but also (complex) services defined in terms of a set of conditions that constrain the order in which service operations should be invoked. Finally, in order to evaluate the developed support platform, we have defined a number of composition scenarios for the (integrated) analysis of gene expression data. The use of the SemanticSCo platform has facilitated the definition of these scenarios, thus allowing the reproduction of the results obtained from different gene expression studies previously documented in the literature.

As pesquisas em biologia molecular têm produzido uma grande quantidade de dados, os quais embutem informações sobre diferentes fenômenos biológicos. Neste sentido, a bioinformática se destaca como uma área de pesquisa multidisciplinar que visa, principalmente, o desenvolvimento de ferramentas (sistemas) computacionais para auxiliar na descoberta de conhecimento a partir de dados biológicos. Dentro da bioinformática, a área de genômica funcional procura estudar as funções gênicas através da medição simultânea e em larga escala dos níveis de expressão gênica de um genoma. Diferentes ferramentas são utilizadas no processo de análise de expressão gênica, cada qual provê suporte a uma atividade de análise específica. Embora alguns ambientes de descoberta de conhecimento ofereçam suporte integrado a este processo de análise e exploração de dados, a maior parte das ferramentas de análise é desenvolvida independentemente de outras ferramentas e ambientes de descoberta de conhecimento. Este cenário representa um desafio para biologistas que precisam combinar e integrar diferentes ferramentas, muitas vezes de forma ad hoc, custosa e sujeita a erros. Modelos conceituais, tais como ontologias, têm contribuído para o sucesso do desenvolvimento de sistemas computacionais em diferentes domínios de aplicação. O desenvolvimento de tais modelos tem por objetivo representar corretamente, em alto nível de abstração, conceitos e situações pertinentes a um dado domínio de interesse. Esta representação abstrata facilita não apenas o entendimento de um dado domínio, mas também serve como base para o processo de desenvolvimento do sistema como um todo. O objetivo deste trabalho é investigar o desenvolvimento e o uso de modelos conceituais em geral e ontologias em particular, na integração de ferramentas na área de análise de expressão gênica. De forma específica, este trabalho tem por objetivo propor uma abordagem para a integração semântica de ferramentas de análise de expressão gênica a partir do uso de conectores e de uma ontologia de domínio. Essa abordagem foi aplicada no desenvolvimento de estudos de caso envolvendo a criação de diferentes ambientes integrados para a análise de expressão gênica e mostrou-se eficaz. / Molecular biology researches are increasingly producing large amounts of data regarding underlying biological phenomena. Bioinformatics is a multidisciplinary research field whose main objective is the development of theories and information systems to help the process of knowledge discovery from biological data. Functional genomics is a field of study bioinformatics concerned with the study of gene function through parallel and large scale expression measurements of a genome. A variety of software tools are usually combined and used in a knowledge discovery process, each providing support for a specific data analysis task. Although some tools are already provided as part of an integrated knowledge discovery environment, most of them are developed independently of other software tools and knowledge discovery environments. This scenario poses a problem and a challenge for biologists that need to combine and integrate different tools in an ad hoc, time consuming and error prone process. Conceptual models, such as ontologies, have contributed to the successful development of information systems in different application domains. The development of such models aims at creating a clear and precise description of the elements of a given domain at a high abstraction level. This abstract and high level description not only promotes a shared understanding of the domain, but also serves as basis for the development process of supporting applications in the domain. This work aims at investigating the development and use of conceptual models in general and ontologies in particular to support the integration of gene expression data analysis systems. Specifically, this work proposes an approach for the semantic integration of gene expression analysis tools using connectors and a domain ontology. This approach was applied in the development of a number of case studies aiming at creating integrated environments for gene expression analysis and proved its effectiveness.

Modernas tecnicas de biologia molecular sao utilizadas para quantificar os niveis de expressao genica de amostras celulares, frequentemente submetidas a diferentes condicoes experimentais. A genomica funcional e a area da bioinformatica responsavel pela caracterizacao do papel funcional dos genes e processos biologicos associados. A analise de expressao genica requer o uso (integrado) de diversas ferramentas de analise para a obtencao de respostas as questoes biologicas de interesse. Uma plataforma de analise pode ser utilizada para facilitar a integracao sintatica e semantica destas ferramentas. Tal plataforma deve oferecer acessibilidade, reprodutibilidade e transparencia, de modo a promover a colaboracao e a disseminacao da pesquisa cientifica. A plataforma SemanticSCo foi concebida para prover suporte a composicao semantica de servicos de analise de expressao genica. Contudo, esta plataforma, em sua versao standalone, apresenta baixa acessibilidade, alem de nao prover suporte para a reproducao e o compartilhamento dos workflows desenvolvidos por seus usuarios. O principal objetivo deste projeto foi ampliar a acessibilidade da plataforma SemanticSCo, bem como introduzir suporte para a reproducao e compartilhamento de workflows. Como resultado principal, desenvolvemos uma nova versao da plataforma SemanticSCo, chamada SemanticSCo Web. A plataforma SemanticSCo Web possui uma interface simples e amigavel, contribuindo para o desenvolvimento colaborativo de pesquisa cientifica reproduzivel no dominio da genomica funcional. Adicionalmente, desenvolvemos uma API chamada SIBRI para permitir a execucao automatica, via framework Activiti, de servicos RESTful modelados como processos BPMN. O uso desta API facilita a integracao de aplicacoes desenvolvidas na linguagem Java e que fazem uso de servicos RESTful, dado que esta API prove suporte a interacao com o usuario em um ambiente web para que o mesmo possa prover valores de parametros que serao utilizados na execucao de um dado servico. Desta forma, a API SIBRI simplifica a incorporacao de novos servicos na plataforma SemanticSCo Web. / Modern molecular biology techniques are used to quantify the levels of gene expression of cellular samples, often submitted to different experimental conditions. Functional genomics is the area of bioinformatics responsible for the characterization of the functional role of genes and associated biological processes. The analysis of gene expression requires the (integrated) use of several analysis tools to obtain answers to a given biological question. An analysis platform can be used to facilitate the syntactic and semantic integration of these tools. Such platform should offer accessibility, reproducibility and transparency in order to promote collaboration and dissemination of scientific research. The SemanticSCo platform was designed to provide support for the semantic composition of gene expression analysis services. However, this platform, in its standalone version, presents low accessibility, and does not provide support for the reproduction and sharing of developed workflows. The main objective of this project was to increase the accessibility of the SemanticSCo platform, as well as to introduce support for the reproduction and sharing of workflows (transparency). As a main result, we have developed a new version of the SemanticSCo platform, called SemanticSCo Web. The SemanticSCo Web platform has a simple and user-friendly interface, contributing to the collaborative development of reproducible scientific research in the functional genomics domain. Additionally, we have developed an API called SIBRI to allow the automatic execution, via Activity framework, of RESTful services modeled as BPMN processes. The use of this API facilitates the integration of applications developed in Java that uses RESTful services, since this API provides support for the interaction with the end user in a web environment so s/he can provide parameters to be used in the execution of a given service. In this way, the API SIBRI simplifies the incorporation of new services in the SemanticSCo Web platform.

Modernas tecnicas de biologia molecular sao utilizadas para quantificar os niveis de expressao genica de amostras celulares, frequentemente submetidas a diferentes condicoes experimentais. A genomica funcional e a area da bioinformatica responsavel pela caracterizacao do papel funcional dos genes e processos biologicos associados. A analise de expressao genica requer o uso (integrado) de diversas ferramentas de analise para a obtencao de respostas as questoes biologicas de interesse. Uma plataforma de analise pode ser utilizada para facilitar a integracao sintatica e semantica destas ferramentas. Tal plataforma deve oferecer acessibilidade, reprodutibilidade e transparencia, de modo a promover a colaboracao e a disseminacao da pesquisa cientifica. A plataforma SemanticSCo foi concebida para prover suporte a composicao semantica de servicos de analise de expressao genica. Contudo, esta plataforma, em sua versao standalone, apresenta baixa acessibilidade, alem de nao prover suporte para a reproducao e o compartilhamento dos workflows desenvolvidos por seus usuarios. O principal objetivo deste projeto foi ampliar a acessibilidade da plataforma SemanticSCo, bem como introduzir suporte para a reproducao e compartilhamento de workflows. Como resultado principal, desenvolvemos uma nova versao da plataforma SemanticSCo, chamada SemanticSCo Web. A plataforma SemanticSCo Web possui uma interface simples e amigavel, contribuindo para o desenvolvimento colaborativo de pesquisa cientifica reproduzivel no dominio da genomica funcional. Adicionalmente, desenvolvemos uma API chamada SIBRI para permitir a execucao automatica, via framework Activiti, de servicos RESTful modelados como processos BPMN. O uso desta API facilita a integracao de aplicacoes desenvolvidas na linguagem Java e que fazem uso de servicos RESTful, dado que esta API prove suporte a interacao com o usuario em um ambiente web para que o mesmo possa prover valores de parametros que serao utilizados na execucao de um dado servico. Desta forma, a API SIBRI simplifica a incorporacao de novos servicos na plataforma SemanticSCo Web. / Modern molecular biology techniques are used to quantify the levels of gene expression of cellular samples, often submitted to different experimental conditions. Functional genomics is the area of bioinformatics responsible for the characterization of the functional role of genes and associated biological processes. The analysis of gene expression requires the (integrated) use of several analysis tools to obtain answers to a given biological question. An analysis platform can be used to facilitate the syntactic and semantic integration of these tools. Such platform should offer accessibility, reproducibility and transparency in order to promote collaboration and dissemination of scientific research. The SemanticSCo platform was designed to provide support for the semantic composition of gene expression analysis services. However, this platform, in its standalone version, presents low accessibility, and does not provide support for the reproduction and sharing of developed workflows. The main objective of this project was to increase the accessibility of the SemanticSCo platform, as well as to introduce support for the reproduction and sharing of workflows (transparency). As a main result, we have developed a new version of the SemanticSCo platform, called SemanticSCo Web. The SemanticSCo Web platform has a simple and user-friendly interface, contributing to the collaborative development of reproducible scientific research in the functional genomics domain. Additionally, we have developed an API called SIBRI to allow the automatic execution, via Activity framework, of RESTful services modeled as BPMN processes. The use of this API facilitates the integration of applications developed in Java that uses RESTful services, since this API provides support for the interaction with the end user in a web environment so s/he can provide parameters to be used in the execution of a given service. In this way, the API SIBRI simplifies the incorporation of new services in the SemanticSCo Web platform.

<p>Over the past decade, the development of macromolecular drugs based on peptides, proteins and nucleic acids has increased the interest in microparticulate drug delivery, i.e., the delivery of drug systems in the nanometer and micrometer ranges. However, little is known so far about the effect that microparticulate systems have on various tissues after administration. Additionally, the knowledge of mechanisms responsible for the uptake and transport of microparticles across the human intestine is incomplete and requires further investigation to improve both the safety profiles and the efficiency of these drug delivery systems.</p><p>This thesis is comprised of two parts. The first one investigates gene expression responses obtained from DNA arrays in local and distal tissues after microparticulate drug delivery. The second part focuses on the mechanisms responsible for the transport of microparticles across epithelial cells lining the intestine.</p><p>The results presented in the first part demonstrated that gene expression analysis offers a detailed picture of the tissue responses after intramuscular or pulmonary administration of microparticulate drug delivery systems compared to the traditional techniques used for such evaluations. In addition, DNA arrays provided a useful and sensitive tool for the initial characterization and evaluation of both local and distal tissue responses, making it possible to distinguish between gene expression patterns related to each studied delivery system.</p><p>The results presented in the second part demonstrated that the surface properties of the microparticle were important for the extent of transport across an <i>in vitro</i> model of the follicle-associated epithelium (FAE), comprised of intestinal epithelial cells specialized in particle transport (M cells). Another important finding was that the enteropathogen bacterium, <i>Yersinia pseudotuberculosis</i>, induced microparticle transport across the normal intestinal epithelium, represented by Caco-2 cells and excised human ileal tissue. This transport was most probably mediated by an increased capacity for macropinocytosis in the epithelial cells.</p>

Over the past decade, the development of macromolecular drugs based on peptides, proteins and nucleic acids has increased the interest in microparticulate drug delivery, i.e., the delivery of drug systems in the nanometer and micrometer ranges. However, little is known so far about the effect that microparticulate systems have on various tissues after administration. Additionally, the knowledge of mechanisms responsible for the uptake and transport of microparticles across the human intestine is incomplete and requires further investigation to improve both the safety profiles and the efficiency of these drug delivery systems. This thesis is comprised of two parts. The first one investigates gene expression responses obtained from DNA arrays in local and distal tissues after microparticulate drug delivery. The second part focuses on the mechanisms responsible for the transport of microparticles across epithelial cells lining the intestine. The results presented in the first part demonstrated that gene expression analysis offers a detailed picture of the tissue responses after intramuscular or pulmonary administration of microparticulate drug delivery systems compared to the traditional techniques used for such evaluations. In addition, DNA arrays provided a useful and sensitive tool for the initial characterization and evaluation of both local and distal tissue responses, making it possible to distinguish between gene expression patterns related to each studied delivery system. The results presented in the second part demonstrated that the surface properties of the microparticle were important for the extent of transport across an in vitro model of the follicle-associated epithelium (FAE), comprised of intestinal epithelial cells specialized in particle transport (M cells). Another important finding was that the enteropathogen bacterium, Yersinia pseudotuberculosis, induced microparticle transport across the normal intestinal epithelium, represented by Caco-2 cells and excised human ileal tissue. This transport was most probably mediated by an increased capacity for macropinocytosis in the epithelial cells.

The formation of blood, hematopoiesis, is a dynamic process originating from a small number of hematopoietic stem cells (HSCs). To sustain hematopoiesis throughout life HSCs have the unique capacity to differentiate into all mature hematopoietic lineages as well as generating more HSCs by a mechanism referred to as self-renewal. However, the regulation of these processes is largely unknown. During embryonic development HSCs expand in the fetal liver, indicating that this environment supports HSC self-renewal. The LIM-homeobox gene Lhx2 is expressed in the fetal liver during this period and Lhx2 null mutant mice die in utero due to severe anemia caused by an environmental defect in the fetal liver. Embryonic stem cells differentiate in vitro, forming embryoid bodies (EBs) containing various tissues including hematopoietic progenitor cells. Introduction of Lhx2 into this system by retroviral transfer led to the generation of cytokine dependent HSC-like cell lines that were multipotent and expressed surface markers similar to embryonic HSCs. However, the specificity and efficiency of this event could not be elucidated. To further evaluate the function of Lhx2 expression during hematopoietic development, Lhx2 was introduced into an ES cell system where expression could be efficiently turned on. This approach revealed that Lhx2 induce self-renewal of distinct multipotent hematopoietic progenitor/stem cells present in the EB, with the ability to form HSC-like cell lines. The Lhx2 induced self-renewal is growth factor specific since stem cell factor and interleukin-6 are necessary and sufficient for this process. However, Lhx2 expression blocked erythroid differentiation and interfered with early ES cell commitment, indicating that the effect of Lhx2 is cell type specific. Since HSCs of early embryonic origin are inefficient in engrafting adult recipients upon transplantation, we wanted to address whether we could generate cell lines retaining this capacity by expression of Lhx2 in hematopoietic cells from adult bone marrow. This led to the generation of clonal and cytokine dependent HSC-like cell lines capable of generating erythroid, myeloid and lymphoid cells upon transplantation into lethally irradiated recipients. When transplanted into stem cell-deficient mice, they contributed to circulating erythrocytes for at least 18 months, revealing a remarkable potential for self-renewal and differentiation in vivo. However, expression of Lhx2 was maintained in vivo and most engrafted mice developed a transplantable myeloproliferative disorder resembling human chronic myeloid leukemia. Thus, elucidation of the mechanism for Lhx2 function in HSC-like cell lines would give insights into both normal and pathological regulation of HSCs. Down-regulation of Lhx2 expression in HSC-like cell lines with inducible Lhx2 expression led to rapid loss of stem cell characteristics and differentiation into various hematopoietic cell types. Thus, global gene expression analysis comparing Lhx2+ HSC-like cell lines to their Lhx2- progeny would give insights into the molecular basis for Lhx2 function in stem cells. A number of differentially expressed genes overlapped with previously reported HSC enriched genes, further emphasizing the resemblance between HSCs and the HSC-like cell lines also at the molecular level. Moreover, a number of genes were identified with functions or expression patterns related to Lhx2 in other organs. Collectively, these data suggest that these HSC-like cell lines represent a relevant model system for normal HSCs on the molecular and the functional level as well as for evaluating Lhx2 function in the development of various tissues in the embryo as well as in disease.

As pesquisas em biologia molecular têm produzido uma grande quantidade de dados, os quais embutem informações sobre diferentes fenômenos biológicos. Neste sentido, a bioinformática se destaca como uma área de pesquisa multidisciplinar que visa, principalmente, o desenvolvimento de ferramentas (sistemas) computacionais para auxiliar na descoberta de conhecimento a partir de dados biológicos. Dentro da bioinformática, a área de genômica funcional procura estudar as funções gênicas através da medição simultânea e em larga escala dos níveis de expressão gênica de um genoma. Diferentes ferramentas são utilizadas no processo de análise de expressão gênica, cada qual provê suporte a uma atividade de análise específica. Embora alguns ambientes de descoberta de conhecimento ofereçam suporte integrado a este processo de análise e exploração de dados, a maior parte das ferramentas de análise é desenvolvida independentemente de outras ferramentas e ambientes de descoberta de conhecimento. Este cenário representa um desafio para biologistas que precisam combinar e integrar diferentes ferramentas, muitas vezes de forma ad hoc, custosa e sujeita a erros. Modelos conceituais, tais como ontologias, têm contribuído para o sucesso do desenvolvimento de sistemas computacionais em diferentes domínios de aplicação. O desenvolvimento de tais modelos tem por objetivo representar corretamente, em alto nível de abstração, conceitos e situações pertinentes a um dado domínio de interesse. Esta representação abstrata facilita não apenas o entendimento de um dado domínio, mas também serve como base para o processo de desenvolvimento do sistema como um todo. O objetivo deste trabalho é investigar o desenvolvimento e o uso de modelos conceituais em geral e ontologias em particular, na integração de ferramentas na área de análise de expressão gênica. De forma específica, este trabalho tem por objetivo propor uma abordagem para a integração semântica de ferramentas de análise de expressão gênica a partir do uso de conectores e de uma ontologia de domínio. Essa abordagem foi aplicada no desenvolvimento de estudos de caso envolvendo a criação de diferentes ambientes integrados para a análise de expressão gênica e mostrou-se eficaz. / Molecular biology researches are increasingly producing large amounts of data regarding underlying biological phenomena. Bioinformatics is a multidisciplinary research field whose main objective is the development of theories and information systems to help the process of knowledge discovery from biological data. Functional genomics is a field of study bioinformatics concerned with the study of gene function through parallel and large scale expression measurements of a genome. A variety of software tools are usually combined and used in a knowledge discovery process, each providing support for a specific data analysis task. Although some tools are already provided as part of an integrated knowledge discovery environment, most of them are developed independently of other software tools and knowledge discovery environments. This scenario poses a problem and a challenge for biologists that need to combine and integrate different tools in an ad hoc, time consuming and error prone process. Conceptual models, such as ontologies, have contributed to the successful development of information systems in different application domains. The development of such models aims at creating a clear and precise description of the elements of a given domain at a high abstraction level. This abstract and high level description not only promotes a shared understanding of the domain, but also serves as basis for the development process of supporting applications in the domain. This work aims at investigating the development and use of conceptual models in general and ontologies in particular to support the integration of gene expression data analysis systems. Specifically, this work proposes an approach for the semantic integration of gene expression analysis tools using connectors and a domain ontology. This approach was applied in the development of a number of case studies aiming at creating integrated environments for gene expression analysis and proved its effectiveness.

Estudos de expressão gênica geralmente envolvem a realização de processos de análise integrados para a obtenção de respostas biológicas de interesse. A realização destes processos frequentemente requer o uso combinado de uma série de ferramentas de software. No entanto, o processo de integração manual de ferramentas pode ser demorado e propenso a erros devido ao crescente número de ferramentas e formatos de dados disponíveis no domínio. De modo a automatizar o processo de integração, algumas abordagens têm sido propostas tanto para a adaptação das ferramentas de análise existentes como serviços web semânticos, quanto para o desenvolvimento de ambientes de suporte à integração (composição) de serviços web semânticos. Embora estas abordagens representem avanços, nenhuma solução adequada para o desenvolvimento e composição de serviços foi especificamente definida para o domínio de genômica funcional. Neste contexto, o principal objetivo deste projeto foi investigar uma solução completa para o desenvolvimento e composição de serviços web semânticos para a análise de expressão gênica. Como parte da solução proposta, definimos uma metodologia integrada para a implementação de serviços web semânticos criados a partir de ferramentas de software existentes e para a anotação semântica destes serviços. Nossa metodologia fornece diretrizes concretas para o desenvolvimento sistemático de serviços, considerando também os principais aspectos técnicos associados ao processo de desenvolvimento. Esta metodologia foi aplicada a um conjunto representativo de serviços que fornecem suporte às principais atividades de análise realizadas em diferentes tipos de dados de expressão gênica. De forma complementar, definimos uma solução completa para a composição semântica de serviços no domínio de análise de expressão gênica. A solução proposta foi implementada em uma plataforma de suporte semi-automático à composição de serviços web semânticos, chamada SemanticSCo. Esta plataforma fornece suporte flexível a todas as atividades envolvidas no processo de composição de serviços, incluindo a criação, publicação, requisição, descoberta, seleção, composição e execução de serviços. Além disto, a plataforma SemanticSCo foi projetada para prover suporte adequado a diferentes tipos de usuários, incluindo biologistas e bioinformatas. Neste sentido, a plataforma fornece aos usuários um alto nível de abstração para a definição de seus processos de análise, permitindo que os mesmos se concentrem mais nas questões de pesquisa biológicas do que nos aspectos subjacentes do processo de composição. Adicionalmente, a plataforma SemanticSCo suporta a definição e incorporação não apenas de serviços simples, definidos em termos de uma única operação, mas também de serviços complexos, definidos em termos de um conjunto de condições que restringem a ordem de invocação de suas operações. Finalmente, de modo a avaliar a plataforma de suporte desenvolvida, definimos diferentes cenários de composição para a análise (integrada) de dados de expressão gênica. O uso da plataforma SemanticSCo facilitou a definição destes cenários, permitindo assim a reprodução dos resultados obtidos a partir de diferentes estudos de expressão gênica previamente documentados na literatura / Gene expression studies usually involve the creation of integrated analysis processes for obtaining responses for a biological question. The creation of such processes often require the combined use of a number of software tools. However, the manual integration of tools can be cumbersome and error prone due to the increasing number of tools and data formats available in the domain. In order to automate the integration process, some approaches have been proposed for the adaptation of existing analysis tools as semantic web services as well as for the development of software environments to support the integration (composition) of semantic web services. Although these approaches present advances, to the best of our knowledge, no suitable solution has been proposed for the development and composition of web services in the functional genomics domain. In this context, this project aimed at investigating a complete solution for the development and composition of semantic web services to support gene expression analysis. As part of the proposed solution, we have defined an integrated methodology for the implementation of semantic web services created from existing software tools and the semantic annotation of such services. Our methodology provides concrete guidelines for the systematic development of services, also taking into account the main technical aspects associated with the development process. This methodology has been applied in the development of a representative set of services that support the main analysis activities performed on different types of gene expression data. Complementary to our methodology, we have defined a complete solution for the semantic composition of web services in the gene expression analysis domain. The proposed solution has been implemented in a software platform to support the semi-automatic composition of semantic web services, named SemanticSCo. This platform provides flexible support to all activities involved in the service composition process including service creation, publication, request, discovery, selection, composition and execution. Additionally, the SemanticSCo platform has been designed to support different types of users, including biologists and bioinformaticians. In this sense, the platform provides users with a high level of abstraction in the definition of their analysis processes, thus allowing them to focus more on biological research issues rather than on underlying details of the composition process. In addition, the SemanticSCo platform supports not only the definition and incorporation of (simple) services defined in terms of a single operation, but also (complex) services defined in terms of a set of conditions that constrain the order in which service operations should be invoked. Finally, in order to evaluate the developed support platform, we have defined a number of composition scenarios for the (integrated) analysis of gene expression data. The use of the SemanticSCo platform has facilitated the definition of these scenarios, thus allowing the reproduction of the results obtained from different gene expression studies previously documented in the literature.