Study On The Degree Distribution Of The Protein-Protein Interaction Networks

With development of system biology experiment,a huge number of protein-protein interactions(PPI) data have been produced.The large-scale PPI networks constituted of the dataset.In 2000,the first network of large-scale protein-protein interactions was accomplished in yeast.It was a historic moment and became another hot spot of protein research.Subsequently,such proteins interaction maps were also brought to success in the fly,worm,coliform and even human.It had great significance to system study behavior of living cell.The protein-protein interaction network is a valuable tool to investigate the molecular networks underlying a living cell.It is widely understood that the degree distribution is one of the most important summary statistic of a network.The accuracy of experimental protein-protein interaction data is often questioned.In this thesis,two formal model selection techniques are used to investigate the stabilities of the degree distributions of four model organism networks in three cases.It is found that the stabilities are only relevant to the size of the subnet in those cases.The result of random sample show that random removal of full network and random sample in difference experiment has little influenced to degree distribution of PPI networks.And the node sampling rates can be below 50%when the degree distributions are stable.Besides,to compare the degree distributions of the union intersection of PPI databases,the result indicated that data of intersection of PPI databases had no effect on degree distribution of PPI network.Specially,in the increase with years case,when the node sampling rate is 6.1%,the degree distribution of the D.melanogaster network is stable.Above all,the thesis also indicated that stretched exponential function gave best description to PPI networks.Analysis result suggests that PPI networks would be more complex than scale-free network,and random removal and false-positive and false-negative would not influence degree distribution of PPI networks stability.