BcePred

BcePred predicts linear B-cell epitopes in protein antigen sequences using optimized combinations of amino acid physico-chemical residue properties.

Key Features:

• Prediction target: Identifies linear B-cell epitope regions within protein antigen sequences.
• Input type: Operates on protein antigen sequences as the primary input.
• Scoring scheme: Applies an empirically optimized composite property scoring scheme based on combinations of amino acid physico-chemical residue properties.
• Residue-property analysis: Benchmarks both individual and combined amino acid physico-chemical residue properties to detect discriminative signals.
• Benchmark dataset: Evaluated against a curated dataset comprising 1,029 non-redundant epitopes from Bcipep and an equal number (1,029) of non-epitopes from SWISS-PROT.
• Performance evaluation: Reports performance across multiple thresholds, with individual-property accuracies of approximately 53–58% and a combined four-property accuracy of ~58.7%.
• Output: Produces predicted linear epitope regions based on composite property scores.

Scientific Applications:

• Epitope prioritization: Prioritizes candidate linear B-cell epitopes for downstream immunogenicity assessment.
• Vaccine design support: Supports selection of peptide targets within vaccine design workflows.
• Complementary analysis: Complements structural, evolutionary, and machine learning–based epitope predictors by providing residue-property–based evidence.
• Baseline evaluation: Serves as a reproducible baseline for evaluating contributions of residue physico-chemical properties to B-cell epitope recognition.

Methodology:

Applies an empirically optimized composite scoring of amino acid physico-chemical residue properties to protein sequences, benchmarks property combinations against 1,029 non-redundant Bcipep epitopes and 1,029 SWISS-PROT non-epitopes, and evaluates performance across multiple score thresholds.