Peptide Resources

AntiVPP 1.0 is an antiviral peptide (AVPs) prediction tool based on Random Forest algorithm, designed to provide efficient computational screening for viral infection therapy.

APD3 is a professional database for antimicrobial peptides (AMPs), first launched in 2003 and expanded from APD2 (2009 version) to the current release. Focusing on natural AMPs with defined sequences and activities, it contains 2,619 peptides, including 261 bacteriocins, 4 archaeal peptides, 7 protist peptides, 13 fungal peptides, 321 plant peptides, and 1,972 animal host defense peptides. Functional categories cover 2,169 antibacterial, 172 antiviral, 105 anti-HIV, 959 antifungal, 80 antiparasitic, and 185 anticancer peptides, with newly added annotations for antibiofilm, antimalarial, insecticidal, wound healing, etc. It supports retrieval by target pathogens, molecular binding partners, post-translational modifications, etc. The amino acid profile analysis module provides key basis for AMP classification, prediction, and design, widely used in research and education.

ArachnoServer is a manually curated database consolidating sequence, structure, function, and pharmacological information of spider venom toxins, focusing on small disulfide-bridged peptides (primary active components). These toxins target neuronal ion channels/receptors and have been developed as pharmacological tools, bioinsecticides, and drug leads. Version 3.0 adds an automated pipeline for detecting toxin transcripts in venom-gland transcriptomes, updates data, enhances mass-based search, and provides detailed "toxin cards" for mature toxins.

AtbPpred is a two-layer machine learning model developed for efficient identification of antitubercular peptides (AtbPs) against Mycobacterium tuberculosis infection. Addressing the need for novel therapies against multidrug-resistant tuberculosis, the model integrates nine sequence feature encodings via a two-step feature selection strategy, using Extremely Randomized Tree (ERT) algorithm in a hierarchical framework: the first layer models each feature set independently, while the second layer fuses probabilities from nine models.

AVCpred is a QSAR-based webserver for predicting antiviral compounds (AVCs), leveraging experimental data from the ChEMBL database against 26 deadly viruses including HIV, HCV, HBV, and HHV.