Breast and Lung Anticancer Peptides Classification Using N-Grams and Ensemble Learning Techniques

Anticancer peptides (ACPs) are short protein sequences; they perform functions like some hormones and enzymes inside the body. The role of any or peptide is related to its structure sequence amino acids that make up it. There 20 types in humans, each them has a particular characteristic according chemical structure. Current machine deep learning models have been used classify ACPs problems. However, these neglected Amino Acid Repeats (AARs) play an essential function peptides. Therefore, this paper, offer promising route for novel anticancer by extracting AARs based on N-Grams k-mers using two peptides’ datasets. These datasets pointed breast lung cancer cells assembled curated manually from Cancer Peptide Protein Database (CancerPPD). Every dataset consists their synthesis activity cell lines. Five different feature selection methods were paper improve classification performance reduce experimental costs. After that, classified four classifiers, namely AdaBoost, Random Forest Tree (RFT), Multi-class Support Vector Machine (SVM), Multi-Layer Perceptron (MLP). classifiers evaluated applying five well-known evaluation metrics. Experimental results showed process provided accurate reached 89.25% 92.56%, respectively. In terms AUC, it 95.35% 96.92% both ACPs, proposed performed competently somewhat equally accuracy, precision, F-measures, recall, except SVM-based selection, which superior performance. As result, significantly improved predictive can effectively distinguish as virtual inactive, moderately active, very active.