Mitigating biomass composition uncertainties in flux balance analysis using ensemble representations

Abstract

29The biomass equation is a critical component in genome-scale metabolic models (GEMs): It is 30 one of most widely used objective functions within constraint-based flux analysis formulation, 31 describing cellular driving force under the growth condition. The equation accounts for the 32 quantities of all known biomass precursors that are required for cell growth. Most often than 33 not, published GEMs have adopted relevant information from other species to derive the 34 biomass equation when any of the macromolecular composition is unavailable. Thus, its 35 validity is still questionable. Here, we investigated the qualitative and quantitative aspects of 36 biomass equations from GEMs of eight different yeast species. Expectedly, most yeast GEMs 37 borrowed macromolecular compositions from the model yeast, Saccharomyces cerevisiae. We 38 further confirmed that the biomass compositions could be markedly different even between 39 phylogenetically closer species and none of the high throughput omics data such as genome, 40 transcriptome and proteome provided a good estimate of relative amino acid abundances. Upon 41 varying the stoichiometric coefficients of biomass components, subsequent flux simulations 42 demonstrated how predicted in silico growth rates change with the carbon substrates used. 43 Furthermore, the internal fluxes through individual reactions are highly sensitive to all 44 components in the biomass equation. Overall, the current analysis clearly highlight that 45 biomass equation need to be carefully drafted from relevant experiments, and the in silico 46 simulation results should be appropriately interpreted to avoid any inaccuracies. 47 1 Introduction 48 Constraint-based modelling methods, such as flux balance analysis (FBA), are popular 49 approaches for analyzing cellular metabolic behaviors in silico (Bordbar et al., 2014). Unlike 50 the kinetic modelling, it does not involve complex kinetic parameters and just requires 51 information on metabolic reaction stoichiometry and mass balances around the metabolites 52 under pseudo-steady state assumption (Lewis et al., 2012). Such simplicity of FBA enabled 53 the development and use of hundreds of genome-scale in silico models for several species 54 across all three domains of life for the study of microbial evolution, metabolic engineering, 55 drug discovery, context-specific analysis of high throughput omics data and for the 56 investigation of cell-cell interactions (Bordbar et al., 2014). 57FBA is an optimization-based approach where a particular objective function is 58 maximized or minimized given a series of constraints such as mass balance, thermodynamic, 59 and enzyme capacity for determining the underlying steady-state fluxes. Among objective 60 functions that have been used to interrogate the metabolic states and cellular behaviors, the 61 maximization of biomass production has been most commonly used in FBA simulations with 62 a principal hypothesis that microbial cells typically strive to grow as fast as possible under 63 expone...

Metabolic network percolation quantifies biosynthetic capabilities across the human oral microbiome

“…Another promising application of our approach is evaluating draft models during the metabolic network reconstruction process. In particular, in building new draft stoichiometric models, the producibility metric, which displays nuanced variability across taxa, could be used as an initial estimate of the biomass composition, to be compared to the reference biomass compositions currently used in most reconstructions (Lakshmanan et al, 2019). More generally, our approach fits into an emerging class of metabolic reconstruction and analysis methods that address uncertainty by statistically sampling ensembles (of environments, as done here; fluxes, as studied extensively [Schellenberger and Palsson, 2009]; or network reconstructions, as recently implemented [Biggs and Papin, 2017; Machado et al, 2018]).…”

A Hybrid Flux Balance Analysis and Machine Learning Pipeline Elucidates Metabolic Adaptation in Cyanobacteria

“…Specifically, when calculating the flux distribution across conditions, biomass was chosen as the primary objective, whereas the secondary objective was set to ATP maintenance, photosystem I, or photosystem II, to reflect the main cellular goals of cyanobacteria. Biomass was chosen as a primary objective to represent the maximization of growth rate and cellular yields ( Feist and Palsson, 2010 ; Yuan et al., 2016 ; Lakshmanan et al., 2019 ), which is a critical consideration for the production of biofuels by cyanobacteria as this informs the substrate uptake rates and maintenance requirements that indicate fundamental cellular growth requirements. The chosen secondary objectives are key pathways involved in energy metabolism during photosynthesis.…”

Integration of Time-Series Transcriptomic Data with Genome-Scale CHO Metabolic Models for mAb Engineering

“…To identify optimal solutions, FBA requires a biologically relevant objective function which represents the metabolic goal of the organism [38], such as maximizing cell growth rate and minimizing total flux. Although biomass production which describes the growth requirements of a cell is widely used in the flux balance analysis [39], several previous studies have demonstrated that flux distributions and predicted cell growth are sensitive to the changes in biomass composition [40,41]. It indicated that accurate representation of biomass is another key to improve the predictive capability of flux balance analysis.…”