Um novo algoritmo de gap filling de rede metabólica aplicando estratégia de expansão de redes

• Autor(a) principal: Colpo, Rodrigo Amarante
• Data de Publicação: 2019
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações do LNCC
• Texto Completo: https://tede.lncc.br/handle/tede/307
• topic: Bioinformática; Simulação (Computadores); Gap Filling; CNPQ::CIENCIAS BIOLOGICAS::GENETICA

Genome-scale metabolic reconstructions are typically created from an annotated genome. However, for not knowing all the functions of the annotated genes, or all the genes in the genome, it is not known all possible reactions that exist in a organism, making metabolic models systematically incomplete. Evidence of this fact is that, if models are created only using an annotated genome, it will be unable to "grow" in conditions where growth was observed in vitro. For the model to "grow" in silico it is necessary to gap fill the model, adding reactions to the model that were not identified during the genome annotation, allowing previously blocked reactions to be activated. This change occurs because the added reactions allowed the substrates of the blocked reactions to be produced, or because they prevented products from the blocked reactions to accumulate inside the network. Genome-scale metabolic models can be semi automatically created using tools that require very little user's experience. However, models created by such tools use metabolites and reactions identifiers that are not compatible with identifiers from KEGG and BiGG Models, what is necessary for the gap filling procedure. Additional problem is the comparison between draft and curated models, since curated models often use BiGG Models identifiers, while automatically created models usually use SEED's. The present work proposes a methodology to translate draft models, from SEED identifiers to BiGG Models identifiers, and translate the KEGG reactions database, from the KEGG identifiers to the BiGG Models identifiers. So that these reactions can be easily used to gap fill the model. Also, based on reactions used by models available in the BiGG Models' database, an interactive menu was created to allow the user to build a customized set of reactions to be used during the gap filling. Finally, all the translated reactions must be stoichiometrically balanced. With the draft model and the reaction dataset created, the gap filling can be performed. The gap filling method developed in this work uses a newly developed method for metabolic network expansion, while more traditional gap filling methods are based on FBA. Therefore, we first studied the necessary conditions for the FBA results and the metabolic networks expansion to be equivalent. With these conditions assured, the method focus on unblocking from the draft model as many original reactions as possible, finding a solution that minimizes the number of reactions added to the model’s final version.