Peptide Resources

AmPEP is a sequence-based predictor for antimicrobial peptides using distribution patterns of amino acid properties and random forest algorithm. It uses a large and diverse dataset of 3268 AMPs and 166791 non-AMPs, and the optimal model with a 1:3 data ratio obtained by 10-fold cross-validation shows 96% accuracy, 0.9 Matthew’s correlation coefficient, and 0.99 AUC-ROC.

amPEPpy is an open-source, multi-threaded command-line tool for predicting antimicrobial peptide sequences from genome-scale data using a random forest classifier, implemented in Python 3 and available on GitHub.

AMPer is an open-source database for antimicrobial peptides (AMPs), using Hidden Markov Models (HMMs) to identify defensins, cathelicidins, cecropins, etc., with 146 mature peptide models and 40 propeptide models achieving 99% cross-validation accuracy. Integrating public resources like Swiss-Prot, HMM scanning identified 229 new AMPs (195 with validated activity), now housing 1,045 mature peptides and 253 propeptides. It supports sequence similarity search and novel AMP discovery, providing computational tools for anti-resistant bacteria drug development.

AmpGram is a novel tool for antimicrobial peptide (AMP) prediction, designed for longer AMPs and high-throughput proteomic screening, outperforming top classifiers like AMPScanner. It uses random forest to identify AMP regions, locates functional AMPs from proteins like lactoferrin and thrombin, provides 10-amino acid fragment predictions, with a web server and CRAN R package.

AMPpred-EL is an effective antimicrobial peptide prediction model based on an ensemble learning strategy, combining LightGBM and logistic regression. Experimental results show that AMPpred-EL outperforms several state-of-the-art methods on benchmark datasets and improves prediction efficiency.