LRC

LRC predicts B-cell epitopes by integrating physicochemical and structural properties of antigen surfaces through logistic regression on graph vertices followed by Markov Cluster Algorithm (MCL) clustering.

Key Features:

• Graph Construction: Constructs a graph representing the antigen surface.
• Logistic Regression Modeling: Applies logistic regression to integrate physicochemical and structural properties and assigns weights to graph vertices.
• Clustering with MCL: Uses the Markov Cluster Algorithm (MCL) to cluster the weighted graph and delineate candidate epitope regions.
• Benchmarking and Validation: Benchmarks predictions on antibody-antigen PDB complexes and compares performance to DiscoTope, SEPPA, and Ellipro using metrics such as sensitivity and specificity.

Scientific Applications:

• B-cell Epitope Prediction: Identifies surface regions on antigens likely to be recognized by antibodies.
• Vaccine Design: Supports selection of antigenic regions for vaccine antigen engineering.
• Drug and Therapeutic Antibody Development: Informs design and optimization of antibody-based therapeutics and related drug development efforts.

Methodology:

Construct a graph from the antigen surface; model physicochemical and structural properties using logistic regression to assign weights to vertices; apply MCL for clustering; evaluate performance on antibody-antigen PDB complexes against DiscoTope, SEPPA, and Ellipro using sensitivity and specificity.