proABC

proABC predicts antibody-antigen contact residues from antibody sequence using the Random Forest algorithm to identify residues that interact with cognate antigens.

Key Features:

• Random Forest Algorithm: Uses the Random Forest machine learning algorithm on sequence-derived features to predict antibody residues contacting antigens.
• Predictive Performance: Reports recall and specificity rates up to 80% for predicted contact residues.
• Sequence Alignment and Annotation: Aligns antibody sequences and annotates source organism and germline family information for downstream analysis.

Scientific Applications:

• Antibody re-engineering: Identification of antigen-contacting residues to inform redesign of antibodies for diagnostic, biotechnological, or therapeutic purposes.
• Molecular docking guidance: Predicted contact residues serve as restraints or guides for molecular docking experiments.
• Study of binding determinants: Analysis of sequence features to determine which antibody sequence elements contribute to antigen binding.

Methodology:

Sequence-based analysis employing the Random Forest algorithm to predict contact residues, together with sequence alignment, germline-family annotation, and analysis of sequence features contributing to binding.