Investigation of Metabolic Modeling Strategies to Predict Phenotypes of Microbial Mutants Under Data Scarcity

Abstract

Phenotype predictions of microorganisms are a developing research area in computational studies. For these predictions, many modeling strategies have been developed; however, most of these strategies are only suitable for wild-type predictions. In this study, because of the lack of methods to predict an accurate mutant phenotype, especially for acetate and biomass yields, the influence of adding protein allocation constraints to Genome-scale Models (GEM) for phenotype predictions of 21 mutant strains of Escherichia coli were studied by combining various modelling strategies available in the literature. Conditional information for only the wild-type or minimal information for the mutants were supplied for the predictions.

Here, we provide combinations of Flux Balance Analysis (FBA), Minimization of Metabolites Balance (MiMBl) and Metabolic Adjustment Minimization Method (MOMA) modeling strategies with Constrained Allocation (CA) and GEM with Enzymatic Constraints using Kinetic and Omics data (GECKO) methods. In order to solve for the optimal flux distribution using these combined strategies, linear (LP) and quadratic programming (QP) were used. Generally, with the combined strategies, the mutant phenotype was predicted with more accuracy compared to using the other strategies in the previous studies, especially for growth rate predictions. According to Pearson Correlation Coefficients (PCC, ρ), when no information about nutrient uptake of mutants were supplied (i.e. under data scarcity), the best acetate yield (ρ=0.53), biomass yield (ρ=0.71) and growth rate (ρ=-0.53) predictions of the experimental data were obtained, using the “GECKO-MiMBl-Reaction” combined algorithm developed in this study. These values were 13 - 64% higher, compared to the PCC values present in the literature for similar scenarios. Overall, with the newly developed algorithms for combined methods using GEM, more accurate predictions for mutant phenotypes of Escherichia coli have been obtained.