Last synchronized: April 13, 2026
FPDB (Functional Peptide Database) is a professional bioinformatics platform that integrates state-of-the-art peptide prediction tools for functional peptide discovery and characterization. All predictors are developed based on the ESM2 protein language model and feature fusion strategies that incorporate sequence-derived embeddings with physicochemical and structural properties. The platform provides comprehensive peptide analysis, including antifungal peptide prediction, anticancer peptide prediction, hemolytic activity prediction, and antimicrobial spectrum prediction.
DL-QSARES is designed to identify antifungal peptides (AFPs) from peptide sequences. This tool employs the ESM2 protein language model to extract deep contextual sequence representations, which are subsequently processed by a Multi-Layer Perceptron (MLP) classifier for prediction. The ESM2-based framework enables accurate identification of antifungal peptides by capturing evolutionary and contextual information embedded within peptide sequences.
DeepACPred is developed for the identification and characterization of anticancer peptides (ACPs). The model integrates deep sequence embeddings generated by ESM2 with physicochemical properties extracted from the AAindex database. By leveraging feature fusion and deep learning strategies, DeepACPred not only distinguishes ACPs from non-ACPs but also predicts the potential cancer types targeted by ACPs through a multi-label classification framework.
Multi-Hemo predicts the hemolytic potential of peptides based on both sequence and structural characteristics. The model combines ESM2-derived embeddings with physicochemical and structural descriptors extracted from the AAindex database. Feature fusion is subsequently performed to capture multiple aspects of peptide properties associated with hemolytic activity.
AMP-SpectraMIC is designed to predict the antimicrobial spectrum of antimicrobial peptides (AMPs). The model integrates ESM2 embeddings with AAindex-derived physicochemical properties to characterize peptide functionality and infer potential antimicrobial activities against different microbial groups.
Sequence Format: Peptide sequences should be provided in FASTA format or entered directly into the sequence input box. Standard single-letter amino acid codes are supported.
Multiple Sequences: Multiple sequences can be submitted simultaneously for batch prediction.
FASTA Files: For large-scale analyses, users are encouraged to submit FASTA files containing multiple peptide sequences. Use the file upload area on the Tools page to upload your FASTA files.
Sequence Length: There are no strict length limitations, but prediction performance is optimized for typical peptide lengths (5–100 amino acids).
Computational Estimates: Prediction results are generated based on trained deep learning models and should be considered computational estimates. Experimental validation is recommended for critical applications.
Large-Scale Analyses: For large-scale analyses, users are encouraged to submit FASTA files containing multiple peptide sequences through the Tools page.
Model Details: Detailed descriptions of model architectures, datasets, evaluation metrics, and performance benchmarks can be found in the respective tool sections above and in the associated publications.
Updates: The prediction models are periodically updated to incorporate new training data and methodological improvements. Check the platform news for the latest updates.
The general dataset constitutes the core of FPDB, featuring exhaustive annotations of functional peptides. These entries are meticulously verified against primary literature sources.
FPDB00005 provides detailed biophysical properties, structural predictions, and biological assay results.
ACCESS_ID Unique system identifier for cross-referencing.
PEPTIDE_NAME Standardized nomenclature used in research.
ORIGIN_SOURCE Taxonomic and environmental source information.
BIO_ACTIVITY Functional tags (e.g., Antibacterial, Antiviral).
The global search bar on the homepage performs a real-time scan across all indexed text fields including FPDB IDs, peptide names, sequences, and biological sources.
Utilize Boolean operators (AND, OR, NOT) in the Advanced Search page to build complex queries. For example, filtering by sequence length AND source organism simultaneously. Visit the Advanced Search page to access multi-criteria filtering.