Proto-oncogene Protein Sequence Classification using RNN and CNN with Attention Mechanism

The fundamental tool for uncontrolled tumour progression is the lack of regulatory capacity of tumour suppression genes (TSG) and mutations in proto-oncogenes (OG). Even though tumour is a diverse complex of several diseases, discovering possibilities of genes connected to OG activity by computational research can aid in the development of medications that specifically target the disease. Attention mechanism in Deep learning has recently become an innovative approach for classifying protein sequences. The attention-based approach can offer a trustworthy and understandable method that aids in overcoming the existing difficulties in describing deep neural networks for classifying protein sequences. In this study, we classify proto-oncogenes (OG) with the help of CNN, Bi_LSTM and Bi_GRUwith attentionmechanisum. of all the three attention mechanisms, Bi_LSTM significantly performs far better than the other two approaches and achives F1-Score upto 97.3% and it is 3% more traditional ML Random Forest approach.