Mining the genome for ‘known unknowns’

• Autor(a) principal: Silva, Nicolas James Queirós da
• Data de Publicação: 2019
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.6/10169
• topic: Algoritmo de Viterbi; Genoma Humano; Open Reading Frame; Pomegranate; Predição de Genes; Domínio/Área Científica::Ciências Médicas::Ciências Biomédicas

Bioinformatics is a multidisciplinary area that combines two major areas: Biology and Computer Science. It’s one of the fastest rising areas of investigation nowadays. It’s also a fundamental area for the processing of data and information from discoveries in the genetics area. One area that is prominent in the bioinformatics area is gene prediction, where various tools are available to aid researchers. Even though there are several gene prediction tools available, the most used are from several years back. They are reliable tools, but need optimization and some are not so flexible for modification. Tools created in the past years base their model on previous tools. In this dissertation work, a new model is proposed. Through ORF extraction from proteincoding sequences of a fasta-formatted file that the user inputs, these are compared to a target sequence of the user’s choice. A profile-HMM is used as the model to compare the sequences, returning a Logp value for each ORF compared with the target sequence. Match, insert and delete state probabilities were modified, to find the best scenario. The Viterbi algorithm was used to train the model, due to its speed. The results obtained were concordant with what we expected: That an ORF, which would be in the target sequence, presented a better Logp value than an ORF from a randomly selected sequence.