Peptide Resources

CAMP is a free online database for antimicrobial peptides (AMPs), manually curating 3,782 sequences divided into experimentally validated (2,766 patent/non-patent) and predicted (1,016) datasets, including source organisms, activity (MIC values), references, and target/non-target organisms. Machine learning-based prediction tools (Random Forests, SVM, Discriminant Analysis) developed from validated data achieve 93.2%/91.5%/87.5% accuracy on test sets, integrated with BLAST and other analysis tools to facilitate sequence-activity/specificity research. 

CancerPPD is a database of experimentally verified anticancer peptides (ACPs) and proteins, manually collected from research articles, patents, and other databases. Current release includes 3,491 ACPs and 121 anticancer proteins, providing comprehensive info on source, activity, terminal modifications, conformation, etc., covering 249 cancer cell lines and 16 assay types. It includes natural/non-natural chemically modified peptides and D-amino acid peptides, with tertiary structures predicted by PEPstr and SMILES-formatted sequences. Integrated tools like keyword search and sequence/structure similarity search support peptide-based anticancer drug design.

CPPsite is a comprehensive database of cell penetrating peptides (CPPs), manually curating 843 experimentally validated CPPs from literature and patents. Each entry provides detailed information including ID, PubMed ID, sequence, chirality, origin, sub-cellular localization, uptake efficiency, mechanism, hydrophobicity, etc. Integrated with secondary/tertiary structure prediction, amino acid composition, and physicochemical property analysis tools, it supports search, browsing, analysis, and visualization mapping, facilitating the development of effective CPP prediction models and drug delivery system research.

ClassAMP is an algorithm based on Random Forests (RFs) and Support Vector Machines (SVMs) for predicting the propensity of a protein sequence to exhibit antibacterial, antifungal, or antiviral activity, facilitating drug discovery programs involving antimicrobial peptides.

con_ACP is a deep learning-based model for anticancer peptide (ACPs) screening, enabling efficient prediction using peptide sequences alone. The model employs contrastive learning to optimize feature representation and replaces traditional data augmentation with two independent encoders, significantly enhancing ACP discriminative ability.