Machine learning models for predicting membranolytic anticancer peptides

After heart disease, cancer is the second leading cause of death worldwide. Recently, membranolytic anticancer peptides (ACPs) have received considerable attention for their ability to target and kill cancer cells. Identification of ACPs is costly and usually time-consuming. Therefore, the development of efficient computational methods is of a great importance to aid in the identification of potential ACP candidates. In the current study, we developed multiple models using support vector machines (SVMs), gradient boosting classifiers (GB), and random forest classifiers (RF) to predict membranolytic anticancer activity given a peptide sequence. Oscillations in physiochemical properties in protein sequences have been shown to be predictive of protein structure and function, and in this work, we are taking advantage of these known periodicities to predict ACP sequences. To this end, Fourier transforms were applied to the property factor vectors to measure the amplitude of the physiochemical oscillations, which served as the features for our models. Peptides targeting breast and lung cancer cells were collected from the CancerPPD database and converted into physiochemical vectors using 10 property factors for the 20 natural amino acids. Using these datasets, cross-validation has been applied to train and tune the models based on multiple training and testing sets. Additionally, feature selection has been performed to further optimize our SVM models. To evaluate the models, performance has been quantified based on cross-validation classification accuracy. Furthermore, to try our prediction accuracy, we have also considered other sets of physiochemical features and properties of amino acids from the literature into our models.