Faculdade de Ciências e Tecnologia (FCT) >
FCT Departamentos >
FCT: Departamento de Química >
FCT: DQ - Dissertações de Mestrado >
Please use this identifier to cite or link to this item:
|Título: ||Hybrid systems biology: application to Escherichia coli|
|Autor: ||Portela, Rui Miguel Correia|
|Orientador: ||Oliveira, Rui|
|Palavras-chave: ||System Biology|
Projection to latent structures
Elementary flux modes
Multiple omic analysis
|Issue Date: ||2011|
|Editora: ||Faculdade de Ciências e Tecnologia|
|Resumo: ||In complex biological systems, it is unlikely that all relevant cellular functions can be fully described either by a mechanistic (parametric) or by a statistic (nonparametric) modelling approach. Quite often, hybrid semiparametric models are the most appropriate to handle such problems. Hybrid semiparametric systems make simultaneous use of the parametric and nonparametric systems analysis paradigms to solve complex problems. The main advantage of the semiparametric over the parametric or nonparametric frameworks lies in that it broadens the knowledge base that can be used to solve a particular problem, thus avoiding reductionism.
In this M.Sc. thesis, a hybrid modelling method was adopted to describe in silico Escherichia coli cells. The method consists in a modified projection to latent structures model that explores elementary flux modes (EFMs) as metabolic network principal components. It maximizes the covariance between measured fluxome and any input “omic” dataset. Additionally this method provides the ranking of EFMs in increasing order of explained flux variance and the identification of correlations between EFMs weighting factors and input variables.
When applied to a subset of E. coli transcriptome, metabolome, proteome and envirome (and combinations thereof) datasets from different E. coli strains (both wild-type and single gene knockout strains) the model is able, in general, to make accurate flux predictions. More particularly, the results show that envirome and the combination of envirome and transcriptome are the most appropriate datasets to make an accurate flux prediction (with 88.5% and 85.2% of explained flux variance in the validation partition, respectively). Furthermore, the correlations between EFMs weighting factors and input variables are consistent with previously described regulatory patterns, supporting the idea that the regulation of metabolic functions is conserved among distinct envirome and genotype variants, denoting a high level of modularity of cellular functions.|
|Descrição: ||Dissertation presented to obtain a Master degree in Biotechnology|
|Appears in Collections:||FCT: DQ - Dissertações de Mestrado|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.