Researches On The Protein-Protein Interaction Prediction Method Based On The Sequence

Bioinformatics focus on the method of analysis and processing biological data. With the integration of molecular biology and computer technology, mathematics, information science and engineering, bioinformatics grows fast. Molecular biology concerns about the activities of life at the molecular level, study of the structure and function of biological molecules and the interaction between DNA, RNA and protein constitute the main research content of molecular biology.Protein is the material basis of life, almost involved in the whole process of life activities. The protein-protein interaction is the interaction between the protein molecules, substantially all of the protein through the interaction with other proteins to accomplish a specific function. Protein interactions play an important role in many cellular activities, such as cell cycle control, protein folding, transcription, translation, and post-translational modification. Protein interactions contribute to a better understanding of protein function.For studying protein interactions, there are two types of methods that is biological experiments and computational methods. Variety calculation methods have been proposed to predict the protein-protein interaction. The method of protein-protein interaction prediction based on the amino acid sequence, does not require much prior knowledge, attracted widespread attention.This paper studies the method to predict protein-protein interactions departure from the sequence. First, introduced the data pre-processing methods, including sequence representation, feature selection methods, and data set constructor. Then use the support vector machine as a learning model, the various factors that affect the protein-protein interaction prediction accuracy is compared. Finally, for the imbalance of the interaction data, we use granular support vector machine, try to solve protein-protein interaction predicting problem. Results of numerical experiment contribute to the study of protein-protein interactions as a reference.