Construction And Analysis Of Biomolecular Networks In Prokaryotes

Biological systems are complex systems,in which all components including nucleic acids,proteins,and small molecules interact with each other,forming complex networks.The usage of network analysis methods in complex theories can help analyze the interactions between components and the structure and acting characteristics of the systems on the whole.The development of high-throughput techno logy and the improvement in computer storing and computing ability have made it possible to construct and analyze molecular networks in genome scale.As a model organism of prokaryotes,Bacillus suzbtilis is more representative and of greater value for basic scientific research.In this work,we constructed the extensive gene regulatory network(GRN)for B.subtilis and analyze it with several complex network methods.With the research going deeper for B.subtilis,the biology data is becoming much more enriched,and there are several important subject databases such as SubtiList,BsubCyc and DBTBS.BsubCyc has integrated multi-aspect information about B.subtilis like genome,proteins,metabolic reactions and regulations,making it a comprehensive database for B.subtilis.But the data obtained officially was stored in flat files which make it difficult to satisfy the individualized requirements.So we designed the MySQL based database according to the original data structure,and imported the data into MySQL database after transformation in the combination use of computer and manual method.Then the MySQL database was further modified and consummated,and finally we got the MySQL biological database of B.subtilis composed of 114 tables,including 20 primary tables and 94 subsidiary tables.The biological database constructed here not only fully reserves the original information,but also is much more efficient in data storage and usage.Using the formerly constructed MySQL biological database and DBTBS as original data source,we extracted and integrated the regulation information including regulator genes,target genes and regulation modes,then constructed the GRN which is comprised of 1515 genes and 2348 regulations.The whole GRN fall into 21 weakly connected components,then we took the biggest component as study object and isolated 6 strongly connected components whose main function is referred to sporulation control.Through the study of cumulative degree distribution,we find out that the total and out-degree distributions are similar to power law distribution,showing the scale-free property,while the in-degree is exponentially distributed.Comparing with random networks in equal scale and three other actual networks,the GRN shows more obvious small-world feature with higher cluster coefficient and lower characterized path length.The hierarchical network model proposed by Barabasi was used to analyze the hierarchical character in GRN by examining the relationship between cluster coefficient C(k)and degree k,after fitting the line we got the power exponent b being very close to the b value equaling to 1 in the complete hierarchical network which indicates that the GRN is a hierarchical network.Then we used the method proposed by Gerstein and classified all genes in the GRN into three layers,the parameter analysis of each layer suggested that genes in middle layer play a bottleneck role in propagating regulatory information among the GRN.At the end of this file,we performed a search of 3-node and 4-node motif among the GRN with FANMOD,and found two kinds of motifs including feed-forward loop and bi-fan.