Computational approaches for the discovery of significant genes in cancer

• Ano de defesa: 2021
• Autor(a) principal: Cutigi, Jorge Francisco
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Abordagem computacional; Bioinformática do câncer; Cancer bioinformatics; Cancer genomics; Cancer mutation data; Computational approach; Dados de mutação em câncer; Gene interaction networks; Genes significativos para o câncer; Genômica do câncer; Mutações significativas para o câncer; Redes de interação gênica; Significant genes in cancer; Significant mutations in cancer
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/55/55134/tde-18082021-100555/

Cancer is a complex disease caused by the accumulation of genetic alterations during the individuals life. These alterations are named genetic mutations, which may be divided into two groups: 1) Passenger mutations: mutations that do not change the behavior of the cell; 2) Driver mutations: significant mutations for cancer, that cause carcinogenesis. Cancer cells have a large number of mutations, in which the large majority of them are passenger, and few mutations are drivers. The identification of significant mutated genes, i.e., genes with driver mutations, is essential for the understanding of the mechanisms of cancer initiation and progression. Such a task is a key challenge in cancer genomics, since several studies have shown many significant genes are mutated at a very low frequency. With the next generation DNA sequencing, large and complex genomic datasets have been generated, creating the challenge of analyzing and interpreting this data. Towards uncovering infrequently mutated genes, gene interaction networks combined with mutation data have been explored. This research presents computational approaches for the discovery of reliable significant cancer genes. Such a genes are prioritized by a network-based method which combines weighted mutation frequency and network neighbors influence, and possible false-positives are detected by machine learning-based method which uses mutation data and gene interaction networks to induce predictive models. An experimental study conducted with six types of cancer revealed the potential of the approaches on the discovering of known and possible novel reliable significant cancer genes.