Abstract
Peptides capable of penetrating the blood-brain barrier (BBB) have shown promise as potential drugs for treating diseases of the central nervous system. Recently, there has been growing interest in studying these BBB peptides (BBPs). In this study, we developed a computational model to effectively distinguish between BBPs and non-BBPs. Our model incorporated three different types of sequence features, and we utilized the least absolute shrinkage and selection operator (LASSO) algorithm to eliminate irrelevant and redundant features. The selected features were then used to train a support vector machine for accurate classification of BBPs and non-BBPs. During the jackknife test, our proposed method achieved classification accuracies of 82.67% and 87.37% on the training and independent testing dataset, respectively. Our approach outperformed state-of-the-art predictors when compared to existing tools used for predicting BBPs.