epitope1D

epitope1D predicts linear B-cell epitopes using explainable machine learning to support vaccine design, immunodiagnostic test development, and antibody production.

Key Features:

• Explainable machine learning: Provides interpretable models that highlight biological features and residue contributions to epitope predictions.
• Graph-based signature representation: Represents protein sequences with a graph-based signature derived from the Cutoff Scanning Matrix algorithm.
• Taxonomy-aware predictions: Incorporates Organism Ontology information to produce taxonomy-aware, organism-specific predictions.
• Two novel descriptors: Integrates two novel descriptors into the predictive framework for enhanced feature representation.
• Robust performance: Achieves Areas Under the ROC curve (AUC) of up to 0.935 in cross-validation and blind tests.
• Comprehensive benchmarking: Evaluated against state-of-the-art methods using distinct benchmark datasets and reported superior performance.

Scientific Applications:

• Vaccine design: Informs selection of linear B-cell epitopes for antigen design and epitope-based vaccine development.
• Immunodiagnostics: Supports development of immunodiagnostic assays by predicting candidate linear B-cell epitopes.
• Antibody production: Guides antibody production by identifying epitope residues and interpretable features relevant for antibody binding.

Methodology:

Integrates two novel descriptors into a machine learning framework, including a graph-based signature derived from the Cutoff Scanning Matrix algorithm and Organism Ontology features, employs explainable machine learning, and evaluates models via cross-validation and blind tests.