Integration of transcriptomics, proteomics and extracellular metabolomics data into flux balance analysis

Transcriptomics, proteomics and extracellular metabolomic data were integrated into flux balance analysis using MARGE function from reframed python package (https://doi.org/10.5281/zenodo.3478380). Flux balance analysis simulations of phenotype predictions were performed following MARGE standard parameter settings. The iMM904 yeast model initially used was transformed to include gene‐protein‐reaction (GPR) associations (Machado et al, 2016). Differences in extracellular metabolites abundances, gene expression levels and protein abundances of metabolic genes were calculated between the following: (i) parental chassis‐derived producers and wild‐type strains; and (ii) evolved and parental chassis‐derived producing strains. Fold changes of differentially (q‐value < 0.1) expressed metabolic genes, protein abundances and extracellular metabolite abundances between two conditions were used in the simulation tool. Relative growth rate differences between strains and wild‐type (Fig 1D) were used to fit “growth_frac” parameter (growth_frac of wild‐type = 1). Flux balance analysis was simulated using the differentially phenotypic changes imposed as lower/upper bounds in the flux of the respective reaction. The IBM ILOG CPLEX Optimizer (version 12.8.0) was used for solving the MILP problems. All simulations were conducted with Python 3.6.9.