Deciphering ambiguous control over fluxes through characterization and reduction of uncertainty

The development of kinetic models is still facing the challenges such as large uncertainties in available data. Uncertainty originating from various sources including metabolite concentration levels, flux values, thermodynamic and kinetic data propagates to the kinetic parameter space and prevents an accurate identification of parameters. The uncertainty in parameters propagates further to the outputs of the corresponding kinetic models, such as control coefficients and time evolutions of metabolic fluxes and concentrations, and it can have deleterious effects on metabolic engineering and synthetic biology decisions. We have introduced a framework recently, iSCHRUNK (in Silico Approach to Characterization and Reduction of Uncertainty in the Kinetic Models), which allows us to characterize uncertainty in the parameters of studied kinetic models and integrate this information into a new population of kinetic models with reduced uncertainty. To this end, iSCHRUNK combines the ORACLE framework with parameter classification techniques. With ORACLE, we first construct a set of large-scale and genome-scale kinetic models that are consistent with the integrated datasets and physicochemical laws. We then employ parameter classification techniques to data-mine the integrated datasets together with the outputs of ORACLE. This way, iSCHRUNK allows us to uncover complex relationships between the integrated data and parameters of kinetic models, and to use the obtained information for constructing an improved set of kinetic models with reduced uncertainty in ORACLE. In this work, we analyzed possible improvements of xylose uptake rate (XTR) in a glucose-xylose co-utilizing S. cerevisiae strain. For this purpose, we computed a population of 200’000 large-scale kinetic models, and we determined the key enzymes controlling XTR. However, large uncertainties due to a limited number of measured flux and metabolite concentration values and lack of data about kinetic parameters led to widespread distributions around zero values for control coefficients of some enzymes such as hexokinase (HXK). We used iSCHRUNK to identify the enzymes and their kinetic parameters that determine negative and positive control of HXK over XTR. Our study showed that by engineering only three saturation states related to two enzymes we could enforce either negative or positive control of HXK over XTR. This result implies that for kinetic modeling of metabolism only a small number of kinetic parameters should be accurately determined and that we can disregard the remaining parameters.

Uncertainty reduction in biochemical kinetic models: Enforcing desired model properties

Vassily Hatzimanikatis, Ljubisa Miskovic, Michaël Roger Germain Moret

A persistent obstacle for constructing kinetic models of metabolism is uncertainty in the kinetic properties of enzymes. Currently, available methods for building kinetic models can cope indirectly with uncertainties by integrating data from different biological levels and origins into models. In this study, we use the recently proposed computational approach iSCHRUNK (in Silico Approach to Characterization and Reduction of Uncertainty in the Kinetic Models), which combines Monte Carlo parameter sampling methods and machine learning techniques, in the context of Bayesian inference. Monte Carlo parameter sampling methods allow us to exploit synergies between different data sources and generate a population of kinetic models that are consistent with the available data and physicochemical laws. The machine learning allows us to data-mine the a priori generated kinetic parameters together with the integrated datasets and derive posterior distributions of kinetic parameters consistent with the observed physiology. In this work, we used iSCHRUNK to address a design question: can we identify which are the kinetic parameters and what are their values that give rise to a desired metabolic behavior? Such information is important for a wide variety of studies ranging from biotechnology to medicine. To illustrate the proposed methodology, we performed Metabolic Control Analysis, computed the flux control coefficients of the xylose uptake (XTR), and identified parameters that ensure a rate improvement of XTR in a glucose-xylose co-utilizing S. cerevisiae strain. Our results indicate that only three kinetic parameters need to be accurately characterized to describe the studied physiology, and ultimately to design and control the desired responses of the metabolism. This framework paves the way for a new generation of methods that will systematically integrate the wealth of available omics data and efficiently extract the information necessary for metabolic engineering and synthetic biology decisions.

2019

Machine Learning for Uncertainty Reduction in Biochemical Kinetic Models

Vassily Hatzimanikatis, Ljubisa Miskovic, Michaël Roger Germain Moret

The primary goal of kinetic models is to capture the systemic properties of the metabolic networks, and we need large-scale kinetic models for reliable in silico analyses of the complex dynamic behavior of metabolism. However, parameter uncertainty hinders the development of kinetic models and uncertainty levels increase with the model size. Current methods for building kinetic models within constraint-based modeling frameworks address uncertainty indirectly by integrating data from different biological origins. We recently proposed iSCHRUNK, a computational approach which combines Monte Carlo sampling methods and machine learning techniques to characterize the uncertainties and to reveal complex relationships between the kinetic parameters and the responses of the metabolic networks. Monte Carlo sampling methods allow us to exploit synergies between different data sources and generate a population of kinetic models that are consistent with the available data and physicochemical laws. The machine learning allows us to data-mine the a priori generated kinetic parameters together with the integrated datasets and derive values of kinetic parameters consistent with the observed physiology. In this work, we modified iSCHRUNK to address a design question: can we identify the kinetic parameters and their values that give rise to a desired metabolic behavior? Such information is important for a wide variety of studies ranging from biotechnology to medicine. As an illustration, we used the proposed methodology to find parameters that ensure a rate improvement of the xylose uptake (XTR) in a glucose-xylose co-utilizing S. cerevisiae strain. Our results indicate that only three kinetic parameters need to be accurately characterized to describe the studied physiology, and ultimately to design and control the desired responses of the metabolism. This framework paves the way for a new generation of methods that will systematically integrate the wealth of omics data and extract the information necessary for metabolic engineering and synthetic biology decisions.

2018

Metabolic control analysis of the central carbon pathway in optimally grown E. coli

Stefano Andreozzi, Vassily Hatzimanikatis, Ljubisa Miskovic, Keng Cher Soh

The engineering of the metabolism requires the building of reliable kinetic models of the metabolic pathways through the integration of the information from heterogeneous data sources. Unfortunately, the quantitative knowledge about fluxes, mechanisms or kinetic parameters of the constituent reactions of the metabolic network is most of the times uncertain or missing, due to limitations in the measurement techniques of the relevant biological properties of the cell and variability in experimental conditions. In this contribution, we employ ORACLE (Optimization and Risk Analysis of Complex Living Entities), the computational framework for the analysis and optimization of metabolic networks that addresses these issues. ORACLE allows us to generate populations of kinetic models that account explicitly for the physico-chemical and thermodynamic features and constraints of metabolic networks. We show how ORACLE incorporates fluxomics and metabolomics data and constraints into the central carbon core model of E. coli to uncover its possible steady state behaviors, when the cells are cultivated under the physiological condition of optimal growth on different carbon sources, both anaerobically and aerobically. Alternative cases of possible intracellular flux and thermodynamic states, corresponding to the given experimental data, are identified and subsequently the kinetic analysis is performed to evaluate how the control over fluxes and concentrations is distributed over the metabolic network. We demonstrate that ORACLE methodology provides guidance for metabolic engineering in spite of uncertain/missing information in the local/global properties of the metabolic network, integrating a statistical account on the semi-quantitative available data with consistent thermodynamic information and plausible hypotheses.

2012

Uncertainty reduction in biochemical kinetic models: Enforcing desired model properties

Vassily Hatzimanikatis, Ljubisa Miskovic, Michaël Roger Germain Moret

2019

Machine Learning for Uncertainty Reduction in Biochemical Kinetic Models

Vassily Hatzimanikatis, Ljubisa Miskovic, Michaël Roger Germain Moret

2018