Rice Tissue Specific Protein Protein Interaction Network Prediction

As the most functional important role in cell,proteins are always interact with each other to achieve the biological process.The protein-protein interaction(PPI)is always an important topic in biological research.As a cereal grain,rice(Oryza Sativa)is the most widely consumed staple food for half of the human population.But until now,we still lack a comprehensive knowledge about rice PPI because there is too many subspecies and lacking experimental data.Nowadays,most PPI database provide data predicted through Interologs and has low prediction power.Therefore it is urgent to build a more reliable rice PPI network.In the other hand,PPI tissue specificity is biologically interesting.So in this study we are going to build a rice tissue specific PPI network by integerating different evidence sources and tissue-spefific gene expression profiles.Firstly,we collected the traning datasets and built a Na?ve Bayesian classifier to predict rice PPI.The classifier is consist of five resources,including from Interologs,shared GO annotation,fusion gene,domaindomain interaction and gene co-occurrence.Through 10-fold crossvalidations,we evaluated the classifier with AUC score 0.974 and get 4380939 interactions with high confidence.Then we defined a rice tissue specific network based on the backgoud PPI network we predicted above.Also we calculated the similarities of specific protein-protein interactions between tissues.Finally,we built a database named RT-Net(http://lilab.life.sjtu.edu.cn:8080/rtnet)to offer the search,display and download of the predicted rice PPI with tissue specific information.