Protein Function Frediction Using Function Associations In Protein–protein Interaction Network

In recent years,the rapid development of high-throughput technology has led to huge amounts of protein–protein interaction(PPI)data and unannotated protein sequences.Many approaches for protein function prediction have been developed which use protein–protein interaction networks information.These approaches make use of the observation that proteins with short distance to each other in a PPI network are more likely to have similar functions.Traditional methods usually only use the dependencies among interacting proteins for each same function.However,not all functions are suitable to predict in this way.The functions which are barely linked with the same function more difficult to predict.The problem of protein function prediction in protein–protein interaction network belong to multi-label collective classification.In multi-label settings,the dependencies among related instances with multiple labels are more complex,rational use of these associations can make up for the shortcomings of traditional methods.The work of this paper mainly includes two aspects: Firstly,to complete the interaction network,we reconstruct the network by adding a number of edges based on protein sequence similarity,and then apply a algorithm to spread function information on interaction network;Secondly,apply a iterative algorithm to predict protein function based on the new network.After reducing occasionality as much as possible,statistical analysis results show that the dependencies among functions based on proteins and interactions actually exist.The proposed method is able to capture the two types of dependencies for protein function prediction.The experimental results show that our algorithm performs better than most of existing protein–protein interaction network based algorithms,adding sequence similarity edges and spread function information can really improve the prediction performance.In addition,the dependencies among functions based on proteins and interactions can be effectively applied to the prediction of protein function.The method proposed in this paper can obtain more effective information from protein-protein interaction networks.