Prediction of Human Genes' Regulatory Functions Based on Proteinprotein Interaction Network

In systems biology, regulatory pathway is one of the most important research areas. However, regulatory pathway is so complicated that we still poorly understand this system. On the other hand, with rapid accumulated information on different organisms, it becomes more and more possible to in-depth investigate regulatory pathway. To understand regulatory pathway well, figuring out the components of each pathway is the most important step. In this study, a network- based method was proposed to classify human genes into corresponding pathways. The information of proteinprotein interactions retrieved from STRING was used to construct a network and jackknife test was employed to evaluate the method. As a result, the first order prediction accuracy was 87.91%, indicating that interactive proteins always have similar biological regulatory functions. By comparing the predicted results obtained from other methods based on blast and amino acid composition, respectively, it implies that our prediction method is quite promising that may provide an opportunity to understand this complicated pathway system well.