Anticancer peptides (ACPs) are becoming more well-known for their ability to target only cancer cells. This ability is more effective than that of many conventional therapies, which often have numerous adverse effects. Even while they show promise, finding new ACPs through laboratory experiments is still expensive and takes a lot of time, which has slowed progress in this area. In this study, we sought to address this issue by developing a machine learning approach that predicts ACPs based on evolutionary information from Position-Specific Scoring Matrices (PSSMs) and key physicochemical properties. To refine the proposed model, we used Support Vector Machine-Recursive Feature Elimination (SVM-RFE) to select the most relevant features, and we compared several classifiers, specifically Support Vector Machines, Random Forest, and Logistic Regression, using 10-fold stratified cross-validation for a robust evaluation. Standard metrics such as accuracy, precision, recall, F1-score, MCC, and ROC-AUC were employed to evaluate predictive performance. From the achieved results, it is evident that combining evolutionary and physicochemical criteria can significantly facilitate the identification of possible ACPs. We have deployed the trained model via a local web interface to enable real-time, sequence-based prediction, facilitating the prompt application of this research. This accessible tool is positioned to catalyze further discovery and innovation in peptide-based cancer treatment development.
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号