Currently, clustering applications use classical methods to partition a set of data (or objects) in a set of meaningful sub-classes, called clusters. A cluster is therefore a collection of objects which are “similar” among them, thus can be treated collectively as one group, and are “dissimilar” to the objects belonging to other clusters. However, there are a number of problems with clustering. Among them, as mentioned in [Datta03], dealing with large number of dimensions and large number of data items can be problematic because of computational time. In this thesis, we investigate all clustering algorithms used in [Datta03] and we present a parallel solution to minimize the computational time. We apply parallel programming techniques to the statistical algorithms as a natural extension to sequential programming technique using R. The proposed parallel model has been tested on a high throughput dataset. It is microarray data on the transcriptional profile during sporulation in budding yeast. It contains more than 6,000 genes. Our evaluation includes clustering algorithm scalability pertaining to datasets with varying dimensions, the speedup factor, and the efficiency of the parallel model over the sequential implementation. Our experiments show that the gene expression data follow the pattern predicted in [Datta03] that is Diana appears to be solid performer also the group means for each cluster coincides with that in [Datta03]. We show that our parallel model is applicable to the clustering algorithms and more useful in applications that deal with high throughput data, such as gene expression data.

No description available.

A enorme quantidade de dados que vem sendo gerada por tecnologias modernas de biologia representam um grande desafio para áreas como a bioinformática. Há uma série de programas disponíveis para a análise destes dados, mas que nem sempre são compreendidos o suficiente para serem corretamente aplicados, ou ainda, há problemas que requerem o desenvolvimento de novas soluções. Neste trabalho, nós apresentamos a análise de dados de duas das principais fontes de dados em larga escala: microarrays e sequenciamento. Na primeira, avaliamos se a estatística do método Rank Products (RP) é adequada para a identificação de genes diferencialmente expressos em estudos de doenças complexas, cujo uma das características é a heterogeneidade genética entre indivíduos com o mesmo fenótipo. Na segunda, desenvolvemos uma ferramenta chamada hunT para buscar por genes alvos do fator de transcrição T - um importante marcador de mesoderma com papel chave no desenvolvimento de vertebrados -, através da identificação de sítios de ligação para o T em suas sequências reguladoras. O desempenho do RP foi testado usando dados simulados e dados reais de um estudo de fissura lábio-palatina não-sindrômica, de autismo e também de um estudo que avalia o efeito da privação do sono em humanos. Nossos resultados mostraram que o RP é uma solução eficiente para detectar genes consistentemente desregulados em somente um subgrupo de pacientes, que esta habilidade é mantida com poucas amostras, mas que o seu desempenho é prejudicado quando são analisados poucos genes. Obtivemos fortes evidências biológicas da eficiência do método nos estudos com dados reais através da identificação de genes e vias previamente associados às doenças e da validação de novos genes candidatos através da técnica de PCR quantitativo em tempo real. Já o programa hunT identificou 4.602 genes de camundongo com o sítio de ligação para o domínio do T, sendo alguns deles já demonstrados experimentalmente. Identificamos 32 destes genes com expressão alterada em um estudo onde avaliamos o transcriptoma da diferenciação in vitro de células tronco embrionárias de camundongo para mesoderma, sugerindo a participação destes genes neste processo sendo regulados pelo T / The large amount of data generated by modern technologies of biology provides a big challenge for areas such as bioinformatics. In order to analyze these data there are several computer programs available; however these are not always well understood enough to be correctly applied. Moreover, there are problems that require the development of new solutions. In this work, we present the data analysis of two main high-throughput data sources: microarrays and sequencing. Firstly, we evaluated whether the statistic of Rank Products method (RP) is suitable for the identification of differentially expressed genes in studies of complex diseases, which are characterized by the vast genetic heterogeneity among the individuals affected. Secondly, we developed a tool named hunT to search for target genes of T transcription factor - an important mesodermal marker that plays a key role in the vertebrate development -, by identifying binding sites for T in their regulatory sequences. The RP performance was tested using both simulated and real data from three different studies: non-syndromic cleft lip and palate, autism and sleep deprivation effect in Humans. Our results have shown that RP is an effective solution for the identification of consistently deregulated genes in a subgroup of patients, this ability is maintained even with few samples, however its performance is impaired when only few genes are analyzed. We have obtained strong biological of effectiveness of the method in the studies with real data by not only identifying genes and pathways previously associated with diseases but also corroborating the behavior of novel candidate genes with the real-time PCR technique. The hunT program has identified 4,602 mouse genes containing the binding site for the T domain, some of which have already been demonstrated experimentally. We identified 32 of these genes with altered expression in a study which evaluated the transcriptome of in vitro differentiation of mouse embryonic stem cells to mesoderm, suggesting the involvement of these genes in this process regulated by T

A enorme quantidade de dados que vem sendo gerada por tecnologias modernas de biologia representam um grande desafio para áreas como a bioinformática. Há uma série de programas disponíveis para a análise destes dados, mas que nem sempre são compreendidos o suficiente para serem corretamente aplicados, ou ainda, há problemas que requerem o desenvolvimento de novas soluções. Neste trabalho, nós apresentamos a análise de dados de duas das principais fontes de dados em larga escala: microarrays e sequenciamento. Na primeira, avaliamos se a estatística do método Rank Products (RP) é adequada para a identificação de genes diferencialmente expressos em estudos de doenças complexas, cujo uma das características é a heterogeneidade genética entre indivíduos com o mesmo fenótipo. Na segunda, desenvolvemos uma ferramenta chamada hunT para buscar por genes alvos do fator de transcrição T - um importante marcador de mesoderma com papel chave no desenvolvimento de vertebrados -, através da identificação de sítios de ligação para o T em suas sequências reguladoras. O desempenho do RP foi testado usando dados simulados e dados reais de um estudo de fissura lábio-palatina não-sindrômica, de autismo e também de um estudo que avalia o efeito da privação do sono em humanos. Nossos resultados mostraram que o RP é uma solução eficiente para detectar genes consistentemente desregulados em somente um subgrupo de pacientes, que esta habilidade é mantida com poucas amostras, mas que o seu desempenho é prejudicado quando são analisados poucos genes. Obtivemos fortes evidências biológicas da eficiência do método nos estudos com dados reais através da identificação de genes e vias previamente associados às doenças e da validação de novos genes candidatos através da técnica de PCR quantitativo em tempo real. Já o programa hunT identificou 4.602 genes de camundongo com o sítio de ligação para o domínio do T, sendo alguns deles já demonstrados experimentalmente. Identificamos 32 destes genes com expressão alterada em um estudo onde avaliamos o transcriptoma da diferenciação in vitro de células tronco embrionárias de camundongo para mesoderma, sugerindo a participação destes genes neste processo sendo regulados pelo T / The large amount of data generated by modern technologies of biology provides a big challenge for areas such as bioinformatics. In order to analyze these data there are several computer programs available; however these are not always well understood enough to be correctly applied. Moreover, there are problems that require the development of new solutions. In this work, we present the data analysis of two main high-throughput data sources: microarrays and sequencing. Firstly, we evaluated whether the statistic of Rank Products method (RP) is suitable for the identification of differentially expressed genes in studies of complex diseases, which are characterized by the vast genetic heterogeneity among the individuals affected. Secondly, we developed a tool named hunT to search for target genes of T transcription factor - an important mesodermal marker that plays a key role in the vertebrate development -, by identifying binding sites for T in their regulatory sequences. The RP performance was tested using both simulated and real data from three different studies: non-syndromic cleft lip and palate, autism and sleep deprivation effect in Humans. Our results have shown that RP is an effective solution for the identification of consistently deregulated genes in a subgroup of patients, this ability is maintained even with few samples, however its performance is impaired when only few genes are analyzed. We have obtained strong biological of effectiveness of the method in the studies with real data by not only identifying genes and pathways previously associated with diseases but also corroborating the behavior of novel candidate genes with the real-time PCR technique. The hunT program has identified 4,602 mouse genes containing the binding site for the T domain, some of which have already been demonstrated experimentally. We identified 32 of these genes with altered expression in a study which evaluated the transcriptome of in vitro differentiation of mouse embryonic stem cells to mesoderm, suggesting the involvement of these genes in this process regulated by T

<p>A numerical simulation case involving space plasma and the evolution of instabilities that generates very fast electrons, i.e. approximately at half of the speed of light, is used as a test bed for scientific visualisation techniques. A visualisation system was developed to provide interactive real-time animation and visualisation of the simulation results. The work focuses on two themes and the integration of them. The first theme is the storage and management of the large data sets produced. The second theme deals with how the Visualisation System and Visual Objects are tailored to efficiently visualise the data at hand. </p><p>The integration of the themes has resulted in an interactive real-time animation and visualisation system which constitutes a very powerful tool for analysis and understanding of the plasma physics processes. The visualisations contained in this work have spawned many new possible research projects and provided insight into previously not fully understood plasma physics phenomena.</p>

A large number and variety of genome-wide genomics and proteomics datasets are now available for model organisms. Each dataset on its own presents a distinct but noisy view of cellular state. However, collectively, these datasets embody a more comprehensive view of cell function. This motivates the prediction of function for uncharacterized genes by combining multiple datasets, in order to exploit the associations between such genes and genes of known function--all in a query-specific fashion. Commonly, heterogeneous datasets are represented as networks in order to facilitate their combination. Here, I show that it is possible to accurately predict gene function in seconds by combining multiple large-scale networks. This facilitates function prediction on-demand, allowing users to take advantage of the persistent improvement and proliferation of genomics and proteomics datasets and continuously make up-to-date predictions for large genomes such as humans. Our algorithm, GeneMANIA, uses constrained linear regression to combine multiple association networks and uses label propagation to make predictions from the combined network. I introduce extensions that result in improved predictions when the number of labeled examples for training is limited, or when an ontological structure describing a hierarchy of gene function categorization scheme is available. Further, motivated by our empirical observations on predicting node labels for general networks, I propose a new label propagation algorithm that exploits common properties of real-world networks to increase both the speed and accuracy of our predictions.

A large number and variety of genome-wide genomics and proteomics datasets are now available for model organisms. Each dataset on its own presents a distinct but noisy view of cellular state. However, collectively, these datasets embody a more comprehensive view of cell function. This motivates the prediction of function for uncharacterized genes by combining multiple datasets, in order to exploit the associations between such genes and genes of known function--all in a query-specific fashion. Commonly, heterogeneous datasets are represented as networks in order to facilitate their combination. Here, I show that it is possible to accurately predict gene function in seconds by combining multiple large-scale networks. This facilitates function prediction on-demand, allowing users to take advantage of the persistent improvement and proliferation of genomics and proteomics datasets and continuously make up-to-date predictions for large genomes such as humans. Our algorithm, GeneMANIA, uses constrained linear regression to combine multiple association networks and uses label propagation to make predictions from the combined network. I introduce extensions that result in improved predictions when the number of labeled examples for training is limited, or when an ontological structure describing a hierarchy of gene function categorization scheme is available. Further, motivated by our empirical observations on predicting node labels for general networks, I propose a new label propagation algorithm that exploits common properties of real-world networks to increase both the speed and accuracy of our predictions.

A numerical simulation case involving space plasma and the evolution of instabilities that generates very fast electrons, i.e. approximately at half of the speed of light, is used as a test bed for scientific visualisation techniques. A visualisation system was developed to provide interactive real-time animation and visualisation of the simulation results. The work focuses on two themes and the integration of them. The first theme is the storage and management of the large data sets produced. The second theme deals with how the Visualisation System and Visual Objects are tailored to efficiently visualise the data at hand. The integration of the themes has resulted in an interactive real-time animation and visualisation system which constitutes a very powerful tool for analysis and understanding of the plasma physics processes. The visualisations contained in this work have spawned many new possible research projects and provided insight into previously not fully understood plasma physics phenomena.

In my thesis, I focused on integrative analysis of high-throughput oncogenomic data. This was done in two parts: In the first part, I describe IntOGen, an integrative data mining tool for the study of cancer. This system collates, annotates, pre-processes and analyzes large-scale data for transcriptomic, copy number aberration and mutational profiling of a large number of tumors in multiple cancer types. All oncogenomic data is annotated with ICD-O terms. We perform analysis at different levels of complexity: at the level of genes, at the level of modules, at the level of studies and finally combination of studies. The results are publicly available in a web service. I also present the Biomart interface of IntOGen for bulk download of data. In the final part, I propose a methodology based on sample-level enrichment analysis to identify patient subgroups from high-throughput profiling of tumors. I also apply this approach to a specific biological problem and characterize properties of worse prognosis tumor in multiple cancer types. This methodology can be used in the translational version of IntOGen.