LSTM_Kmal: Prediction of Malonylation Based on LSTM and Feature Confusion

Lysine malonylation(Kmal) is one of the most important Post-translational modifications(PTMs), which plays significant roles in various cellular functions. Therefore, accurate identification of protein Kmal is important to reveal its molecular function associated with many diseases. However, traditional biological experimental technology is always time-consuming and expensive shortcomings, computational-based methods have gained more and more attentions. In this study, we proposed a new model named LSTM_Kmal to identify Kmal and non-Kmal sequences based on the Long short-term memory (LSTM). Here, five features encoding methods were used to characterize the proteins, and the LSTM was utilized to construct the classification. The experimental results show that LSTM_Kmal has good robustness and generalization, with accuracy values of 91.94%, 90.44%, and 93.25% on H.sapiens, E.coli, M.musculus, respectively.