Filtragem robusta de SNPs utilizando redes neurais em DNA genômico completo

Abstract

With the growing advances in genomic sequencing platforms, new developments on computational models are crucial to analyze, effectively, the large volume of data available. One of the main complexities, variations and peculiarities of a genome are single nucleotide polymorphisms (SNPs). The SNPs, which can be found in the genome of isolated individuals or groups of individuals of a specific population, are originated from inserts, removals or substitutions of bases. Single nucleotide variation, such as SNPs, can modify the production of a protein. Combination of all such modifications tend to determine variations on individuals characteristics of the specie. Thus, this phenomenon usually produces functional or phenotypic changes which, in turn, can result in evolutionary consequences for individuals with expressed SNPs. Among the numerous challenges in bioinformatics, the discovery and filtering of SNPs in genomic DNA is considered an important steps of the genome assembling post-processing. This dissertation has proposed and developed a computational method able to filtering SNPs in genome, using the genome assembled from sequences obtained by new generation platforms. The computational model presented is based on machine learning and computational intelligence techniques, aiming to obtain an efficient filter to sort SNPs in the genome of an individual, regardless of the sequencing platform adopted.