Prediction And Analysis Of Multiple Substrates Utilization And Gene Knockouts Based On Metabolic Models With Enzyme Constraints

GEMs(Genome-scale metabolic models)are widely used to calculate metabolic phenotypes.The method enhances a GEM with enzymatic constraints(EC)is one of the most effective extensions of GEMs,which can improve the prediction ability of GEMs.At present,the application analysis using EC models is limited,and there is no analysis method based on EC models.In this study,we developed an analysis process about the prediction of the growth strategy of microbes on mixed carbon sources based on EC models,and did predictive analysis.Besides,we developed a predictive algorithm for cell metabolic state after gene knockout based on EC models.Firstly,we designed a set of analysis flow for the prediction of the growth strategy of microbes on mixed carbon sources,and developed a minimum enzyme cost algorithm suitable for EC models for the realization of the flow.We introduced the concept of pathway efficiency,and calculated the pathway efficiency using the minimum enzyme cost algorithm,so as to predict the cell’s mixed substrate utilization strategy by comparing the pathway efficiency.We analyzed 55 two-substrate mixtures of 11 carbon sources using minimum enzyme cost algorithm and eco ECM,and the results showed that 9 groups were diauxie and the remaining 46 groups were coutilization.Compared with the 26 experimental studies investigated,only 1 group was inconsistent with the experimental study.A prediction of the mutant E.coli was made and the prediction results were consistent with the experimental study.This is the first time to simulate this phenomenon in EC models,which provides a new scenario for the application and extension of EC model.Secondly,we learned the concept of MOMA(minimization of metabolic adjustment),developed the algorithm MOEA(minimization of enzymatic adjustment),which was used to predict the metabolic state of cells after gene knockout based on EC models.We used eco ECM for our research,for the prediction of cell growth after nonisozyme gene knockout,the SSE(sum of error squares)of MOEA was 0.068,which was 85.4% and 87.4% lower than that using EC models and MOMA respectively.For the prediction of internal metabolic fluxes,the average correlation coefficients predicted by MOEA was 0.94,which was 12% and 7% higher than those predicted by MOMA and EC models.MOEA can achieve the analysis of isozyme gene knockout that MOMA cannot simulate.As to the prediction of cell growth after isozyme gene knockout,the SSE predicted by MOEA was 0.16,whitch was 26% lower than that predicted by using EC models directly.MOEA is the first algorithm based on EC models to predict cell metabolic state after gene knockout,which is expected to provide guidance for metabolic engineering.