Low Complexity Classification of Thermophilic Protein using One Hot Encoding as Protein Representation

The laborious, and cost-inefficient biochemical methods for identifying thermophilic proteins necessarily require a rapid accurate method proteins. Recently, machine learning has become more effective specific classes of extremophiles. There is still need low-cost proteins, despite the fact that studies employing yielded superior results to conventional methods. Here, we avoid problem manually crafted features, which involves experts defining extracting set features using only protein sequences as input various computational This study classifies their counterparts in one-hot encoding representation bidirectional long short-term memory (BiLSTM) model. model achieved an accuracy 92.34 percent, specificity 91 sensitivity 93.77 other models reported elsewhere rely on number features. In addition, trustworthy objective data independent evaluation make this competitive with other, models.