Peptide Resources

ACPred is a bioinformatics tool for the prediction and characterization of anticancer peptides (ACPs), developed by utilizing powerful machine learning models (support vector machine and random forest) and various classes of peptide features. It is observed by a jackknife cross-validation test that ACPred can achieve an overall accuracy of 95.61% in identifying ACPs.

ACPred-FL is a webserver for predicting anticancer peptides (ACPs) based on sequence information, using feature representation learning to extract discriminative features from SVM-based models, combined with two-step feature selection to generate a five-dimensional feature vector for accurate ACP identification.

ACPred-Fuse is a machine-learning based predictor for anticancer peptides, integrating 29 sequence feature descriptors to build a feature representation model. It fuses class information, probabilistic features, and handcrafted sequence features, using random forest for automatic and accurate prediction of anticancer activity in protein sequences.

ACPScanner is the first tool for specific anticancer peptide (ACP) activity prediction via integrated machine learning, adopting a two-level prediction architecture: first distinguishing ACPs from non-ACPs, then predicting specific activity types for potential ACPs. It fuses sequence features, physicochemical properties, secondary structure information, and deep learning embedding features, combining customized deep learning and statistical learning models to build feature space, demonstrating superior prediction performance in independent tests compared to existing methods, with a web server and open-source code/datasets available.

ACVPred is an algorithm for predicting anti-coronavirus peptides (ACVPs), designed to address data scarcity and interpretability challenges in COVID-19 therapeutic research. It innovatively integrates two few-shot learning strategies: transfer learning from pre-trained Transformer models to transfer sequence representation capabilities, and data augmentation via oversampling to expand limited ACVP datasets, effectively enhancing model generalization in small-sample scenarios.