Recently, protein-antibody therapeutics becomes a hot search topic. In this paper, all protein interaction data files are collected from INTERPARE. Protein sequence and its secondary structure both are used to build HMM mathematical model. We randomly take 80% data to train positive and negative HMM models and 20% data to test. The accuracy of this approach can reach to 79.80%. This model can further be used to predict protein function sites and predict if a protein interacts with other proteins.