ISBN: 978-981-14-4787-7 DOI: 10.18178/wcse.2020.06.003
Deep learning for Protein-Protein Interactions Predication
Abstract— As the main component, proteins play an important role in the cell activities of organisms. Most organisms carry out the cell activities by protein-protein interactions (PPI), so deep researches about PPI is of great significance. Some machine learning methods have been applied to predict PPI by extracting features from massive protein data and then training the models to implement classification. However, these methods can only be used to deal with balanced datasets and their effects can be improved further. We proposed a deep learning method based on Bi-LSTM model to predict PPI. For protein sequences, our method automatically encoded the amino acids and represented the protein sequences, and then extracted the sequence features and implemented classification. Experimental results showed that our method can achieve higher accuracy than advanced methods, and can solve the problem of unbalanced datasets.
Index Terms— protein-protein interactions (PPI), protein sequences, Bi-LSTM
Shengyu Lu, Beizhan Wang, Hongji Wang
School of Informatics, Xiamen University ,CHINA
Cite: Shengyu Lu, Beizhan Wang, Hongji Wang, "Deep learning for Protein-Protein Interactions Predication " Proceedings of 2020 the 10th International Workshop on Computer Science and Engineering (WCSE 2020), pp. 13-18, Shanghai, China, 19-21 June, 2020.