SPACE2

SPACE2 performs structure-based clustering of antibodies using machine learning-driven structure prediction to identify antibody groups that bind common epitopes at epitope-level resolution.

Key Features:

• Machine Learning-Based Structure Prediction: Uses state-of-the-art machine learning models for antibody structure prediction, addressing limitations of template-based modeling.
• Novel Clustering Protocol: Clusters antibodies by structural characteristics of binding sites rather than by sequence identity, enabling identification of functionally consistent, sequence-diverse clusters.
• High Dataset Coverage and Diversity: Benchmarked across six diverse antigen-specific antibody datasets to achieve higher dataset coverage and identify clusters diverse in sequence, genetic lineage, and species origin.
• Epitope-Level Resolution: Provides epitope-level resolution for detailed insights into antibody–antigen binding interactions.

Scientific Applications:

• Antibody Research: Identify and characterize antibodies that bind the same epitopes across diverse sources, aiding vaccine development, therapeutic antibody discovery, and immunological research.
• Structural Biology: Provide structural insights that complement sequence-based analyses to enhance understanding of antibody–antigen interactions.

Methodology:

SPACE2 integrates machine learning-driven structure prediction with a clustering protocol that groups antibodies by structural characteristics of their binding sites rather than by sequence identity.