EpitopeVec

EpitopeVec predicts linear B-cell epitopes from protein sequences to support peptide vaccine design, immuno-diagnostic reagent development, and antibody production.

Key Features:

• Input format: Accepts protein sequences in FASTA format.
• Output: Identifies candidate peptides of specified amino acid lengths and reports prediction probabilities for each peptide.
• Feature integration: Combines residue properties, modified antigenicity scales, and protein language model-based representations (protein vectors) for prediction.
• Species-specific optimization: Provides a specialized model trained on a large viral dataset for improved viral linear BCE prediction.
• Performance and generalizability: Demonstrates superior accuracy and area under the curve (AUC) in benchmarking against state-of-the-art methods and addresses generalizability issues where existing methods show cross-testing accuracies of ~51–53%.

Scientific Applications:

• Vaccine design: Supports identification of linear B-cell epitope candidates for peptide vaccine development.
• Immunodiagnostics and antibody production: Facilitates selection of peptides for immuno-diagnostic reagent design and antibody generation.
• In silico epitope discovery: Prioritizes candidate BCEs for downstream experimental validation in immunology and infectious disease research.

Methodology:

Accepts FASTA protein sequences and predicts linear BCEs by integrating residue properties, modified antigenicity scales, and protein language model-based representations (protein vectors), outputs peptides with prediction probabilities, includes a model trained on a large viral dataset, and was benchmarked against state-of-the-art methods using accuracy and AUC (cross-testing baseline ~51–53%).