Leveraging Deep Learning for Classification and Prediction of Antimicrobial Peptides

In our study, we delve into the pressing issue of antimicrobial resistance, recognizing its significance in today’s world. We focus on the crucial task of efficiently classifying Antimicrobial Peptides (AMPs), which is pivotal for advancing drug discovery and therapeutic development. Our research compares two distinct approaches for AMP classification, each with its merits. The First Approach employs KD scores as a basis for classification and utilizes an optimized deep learning model, achieving an accuracy of 93%. Subsequently, we introduce an enhanced classification model named AMP-XLNet. Leveraging XLNet, a cutting-edge deep learning architecture, we aim to extract structural features from input peptides to discern whether they are AMPs or not. Our comparative analysis demonstrates the superiority of our proposed AMP-XLNet model over other deep learning-based methods. With a test accuracy of 96.344%, an F1 score of 96.354%, precision reaching 96.08%, and recall at 96.62%,