Multiview Learning and Decision Fusion Based Ensemble to Predict Anti-Cancer Peptide

The spread of a life-threatening disease such as cancer presents a significant threat to the entire global population. The disease in question caused 10 million deaths worldwide in 2020, which has motivated us to explore innovative methods for treating cancer. Although numerous pharmaceutical solutions are accessible, they often have a hefty price tag and can adversely affect human cells. Anticancer peptides are a recently emerging field in cancer diagnosis, as they have a low likelihood of causing adverse reactions. The current study proposed using a multiview and decision fusion approach with stacking ensemble learning to predict anticancer peptides. Multiview learning and decision fusion is an emerging deep learning technique that enhances performance by integrating information from multiple views. Finally, the present work uses three 1-D conventional neural networks as the base classifier and a neural network as the meta-classifier to create a stacking ensemble for anticancer peptide prediction. The proposed ensemble scores a classification accuracy of 95.38%, an MCC value of 0.843, and an AUC score of 0.961 on a standard ENNAACT anticancer dataset, surpassing the performance of state-of-the-art approaches.