StarPepWeb

StarPepWeb democratizes access to StarPepDB, one of the largest curated repositories of bioactive peptides, by integrating non-redundant sequences from multiple public databases into a source-tracked graph enriched with metadata, physicochemical features, and predicted 3D structures for integrative bioactive peptide analysis.

Key Features:

• Database Integration: Integrates 45,120 non-redundant bioactive peptide sequences from 40 public databases with source tracking.
• Graph-Based Representation: Represents peptides in a graph enriched with metadata, physicochemical features, and predicted 3D structures from ESMFold.
• Peptide Descriptors: Represents each peptide with ESM-2 embeddings and iFeature descriptors.
• Metadata-Aware Filtering: Supports filtering based on metadata attributes.
• Similarity Search: Performs alignment-based similarity searches with single and multiple queries.
• Interactive Visualization: Provides interactive visualization of peptide data and relationships.
• Reproducible Downloads: Offers reproducible versioned downloads including Neo4j exports.

Scientific Applications:

• Bioactive Peptide Exploration: Enables comprehensive and reproducible exploration of the bioactive peptide landscape across diverse functional, taxonomic, and structural classes.
• Peptide Similarity Analysis: Supports comparison of bioactive peptides through alignment-based similarity searches using single or multiple query sequences.
• Structural Peptide Analysis: Facilitates investigation of peptide structures through ESMFold-predicted 3D models and associated physicochemical features.

Methodology:

StarPepWeb employs ESM-2 for generating peptide embeddings, iFeature for computing peptide descriptors, and ESMFold for predicting 3D structures. The platform uses a microservice-oriented architecture and implements alignment-based similarity search algorithms.