Research On The Auto-reconstruction And Analysis Of Genome-scale Metabolic Network Model

With the completion of the genome sequencing and generation of large amounts of data,Genome Scale Metabolic Model has been become an indispensable research tool for systemsbiology. However, the speed of model construction is far less than the speed of genomesequencing. One of the reasons for this is that we understood poorly for the physiological andbiochemical mechanisms of many species. Another important reason is that the metabolicnetwork reconstruction process requires a lot of manual work and very time-consuming.Therefore, the auto-reconstruction of metabolic networks becomesthe major issues to promotethe development of the metabolic network.Inaddition, genome minimization has become oneof the focus researches on synthetic biology. It can provide the ideal chassis cell for theapplication of biotechnology. Through the genome minimization simulation and analysis withthe help of information technology, we can get a predictable metabolic network whichcontainsthe most simple genes under normal growth. It also can provide guidance for thedesign of experiments to obtain the simplest genome. As an illustration example, the allautomatic methods were implemented in the process of genome-scale metabolicreconstruction of Pichia stipitis. The main findings are as follows:(1) To build Pichia stipitis CBS6054genome-scale metabolic network model forinstance, the paper has been studied for the auto-reconstruction of metabolic network whichbased on Java and Perl as the assistance language. Three kinds of methods for reconstructionof genome-scale metabolic network, which were based on KEGG online database,Uniprot-MetaCyc databases and homologous alignment, have been studied for the processautomation. Meanwhile, it has proposed an algorithm called Mahalanobis distance which wasused to calculate the distance between reactions character frequency histogram. Thisalgorithm can be used to the auto-integration of the draft model reactions and identify the corebreakpoint. All above of these can improve the efficiency of model reconstruction.(2) Supplemented literature data to the model which based on the auto-integration, andthen debugged and corrected the converted mathematical model with the help of COBRAToolbox in Matlab. We finally got a P.stipitis CBS6054genome-scale metabolic networkmodel iWC978. The model contains1585reactions,1145metabolites and978genes. And allreactions are located in eight compartment (mitochondria, peroxisomes, endoplasmicreticulum, Golgi apparatus, vacuole, nucleus, cytoplasm and extracellular cell). The coveragerate of annotated open reading frame of functional genomics research16.8%. The model isdivided into62metabolic pathways which are corresponding to eight cell subsystems.(3) The model data were simulated growth phenotype including carbon utilization andethanol fermentation. FBA results showed that model can growth on carbon sources (glucose,xylose, rhamnose, cellobiose, D-mannose, L-arabinose, D-galactose and xylitol) and canfermente glucose, xylose, cellobiose, D-mannose and D-galactose to producte ethanol undercontrolled oxygen conditions. This result is in line with the literature and accordance for thephysiological characteristics of P.stipitis. Thus illustrating the effectiveness of the model.(4) Accordance with the "top-down" research ideas based on the model iWC978, using amethod of multi-objective genetic algorithms and by designing the program, We finally got and analyzed the P.stipitis minimized metabolic network models, which can growth onglucose.