Genome-scale Metabolic Network Reconstruction Of Bacillus Subtilis168

Network reconstruction is the core of Systems Biology, the hub which linksvarious “omics” data, high-throughput information to bio-process modification,medical and drug study, environment protection and new resources development, thetool of enhancing the efficiency of metabolic engineering, genetic engineering andenzymitic kinetics. Among network reconstructions, genome-scale metabolic networkreconstruction is the one that deeply studied. Till now, there are altogether87genome-scale models for61organisms, which have already been applied in the fieldof metabolic engineering, drug discovery, phenotype prediction and networkproperties analysis. Meanwhile, as a mode bacterium out of industrial bacteria,B.subtilis has broad usages and representativeness. Thus, it has wide applicable valuesand perspectives to reconstruct the genome-sacle metabolic network of this bacterium.In this paper, the reconstruction starts from genome annotation and is mainlybased on manual curation literature and experimental data. COBRA Toolbox initiatedin the MATLAB is used for simulation, while glpk is employed as the linearprogramming solver. All the computation is based on FBA; the model is modifieduntil it can compute correctly. Phentype prediction and analysis is done using thecurated model.The genome-scale metatolic model for B. subtilis contains1141genes,1720reactions (including244flux exchange reactions) and1441metabolites including1197intracellular metabolites and244extracellular metabolites. The model can utilizesingle carbon source and multiple carbon sources, and can compute growth rate andribolfavin production rate correctly under certain carbon source uptake rate. The totalaccuracy of single gene deletion study is95.2%(1086/1141) using LB medium, andthe essential and nonessential gene prediction accuracy are86.1%（192/223）and97.4%（894/918）respectively. The total accuracy of substates utilization is73.4%(199/271), with utilizable and unutilizable substates predition accuracy being67.2%（123/183）and86.4%（76/88）respectively.