ABSTRACT
Nesta tese são apresentados dois estudos in silico de conjuntos de dados de expressão. Ambos os estudos originaram-se de conjuntos de dados de expressão feitas de "Open Reading Frame ESTs" ou ORESTES. Estes ORESTES foram gerados no projeto genoma humano de câncer da FAPESP/LICR. A ênfase da primeira parte da tese está no estudo de expressão diferencial de genes em tumor de mama, enquanto a segunda parte da tese está no estudo da expressão de novos genes em câncer colorretal. A combinação do uso de ORESTES e a informação disponível dos bancos de dados de UniGene e SAGE caracterizaram o transcriptoma de células normais e tumorais de mama. Neste estudo, identificamos 154 genes como candidatos de genes que são super-expressos em células tumorais de mama... Neste trabalho demonstramos que a metodologia ORESTES pode contribuir para a descoberta de novos genes. Em câncer colorretal observa-se um tipo específico de instabilidade genômica, caracterizado por alterações no tamanho das unidades simples de seqüência repetitiva ou microsatélites. Muitos dos transcritos de baixa abundância podem ser essenciais para determinar fenótipos celulares normais e patológicos e podem ser responsáveis pelas diferenças fundamentais poucos compreendidas entre os diferentes fenótipos de câncer colorretal... As análises aqui apresentadas poderão contribuir para o entendimento das diferenças fundamentais em características clínicas, patológicas e moleculares dos cânceres coloretais com estabilidade (MSS) e instabilidade (MSI) de microsatélites. Com a abordagem computacional apresentada, observamos que a metodologia ORESTES pode ser complementar a outras tecnologias de larga escala de expressão gênica (SAGE, bibliotecas normalizadas de ESTs) na identificação de genes novos com importantes papéis em tumorigênese...
In this thesis, two in silico studies of expression datasets are presented. Both studies start with datasets of ORESTES or "Open Reading Frame Expressed Sequence Tags". These ORESTES were produced within the human cancer genome project of the FAPESP I LICR. The emphasis of the first part of the thesis is on the differential expression of genes in breast tumors. The second part of the thesis emphasizes the study o f novel genes in colorectal cancer. The combination o f the use of ORESTES and the publicly available information in the databases o f Uni Gene and SAGE lead to the characterization o f the transcriptome o f normal and tumour breast cells. In this study, we identified 154 genes as candidate up-regulated genes in breast tumour cells. Among these, 28 genes have been shown by others to be overexpressed in breast or other tumours. Using RT-PCR, we tested 11 candidate genes and found that 9 were indeed overexpressed in breast tumour cells. Furthermore, 99 genes were validated in silico by SAGE data. Of the 55 genes that were not confirmed by SAGE, 42 have their corresponding cluster composed solely by ESTs. These 42 clusters have no functional annotation and the function of these genes is unknown. The transcripts of the genes that are represented by these 42 clusters are likely to be expressed at low leveis in breast tissue. These results led to a more profound study of low abundance genes among which probably most of the novel genes can be found. As the majority of novel genes are expressed at low leveis, difficulties are encountered in identifying them with gene expression techniques like SAGE (Serial Analysis of Gene Expression) or EST (Expressed Sequence Tag) libraries. In this work we show the contribution of the ORESTES methodology in identifying novel genes. In colorectal cancer, a specific type of genetic instability characterized by length alterations within simple repeated sequences, termed microsatellite instability (MSI) is seen in the majority of hereditary nonpolyposis colorectal cancers (HNPCCs) and in a subset o f sporadic cancers. Many o f the low abundance transcripts could distinguish between different phases of tumorigenesis. The analyses presented in this work could contribute to the understanding of fundamental clinicai, pathological and molecular differences between colorectal cancers with stability in microsatellites and colorectal cancers with instability in microsatellites. With the described computational approach, we observed that the ORESTES methodology could be complementary to other large scale gene expression technologies (SAGE, normalized EST libraries) in the identification of novel genes with important roles in tumorigenesis (AU)