Prediction Analysis of Mycobacterium Tuberculosis by using Deep Learning Model

It will be helpful for the biologists to predict Mycobacterium tuberculosis epitopes accurately. Prediction of these epitopes will save time, money, resources and human efforts. The prediction of Mycobacterium tuberculosis epitopes are beneficial while designing drugs and antibotics to treat this disease. There are a few machine learning models exists which is doing such predictions. Thus, in the proposed study, we have analysed that the deep learning model is performing outstanding as compare to the machine learning models. These models are used to classifiy that the given peptide is epitope or nonepitope. The proposed model has the capacity to predict variable length peptide. The primay focus is to identify B cell or T cell epitopes in focused antigen while designing antibody, vaccines and immune-diagnostic tests. So, it is the urge to find an efficient method which can be used to treat mycobacterium tuberculosis. In the proposed study, deep learning model has been trained to forecast mycobacterium tuberculosis epitopes of variable length. The feature selection technique, relevant features have been filtered out and used to train the models. The proposed model has achieved accuracy 89.0%, specificity 0.89, sensitivity 0.80. Also, repeated 5-fold cross-validation has been carried out to check the consistency of model.