AB-Amy

AB-Amy predicts the amyloidogenic risk of therapeutic antibody light chains to assess aggregation-associated safety risks in antibody development.

Key Features:

• Machine Learning Model: Employs a support vector machine (SVM) to classify amyloidogenic versus non-amyloidogenic antibody light chains.
• Feature Representation: Uses dipeptide composition as the primary input feature linked to protein aggregation propensity.
• Training Dataset: Trained on a dataset comprising 742 amyloidogenic and 712 non-amyloidogenic antibody light chains.
• Performance: Achieves an area under the ROC curve (AUC) of 0.9651 for distinguishing amyloidogenic from non-amyloidogenic light chains.
• Early Screening Capability: Enables in silico early screening of antibody candidates for aggregation-related risk.

Scientific Applications:

• Antibody Development: Supports selection and optimization of therapeutic antibodies by predicting light-chain amyloidogenicity.
• Safety Assessment: Identifies potential aggregation-related safety risks during preclinical candidate selection.
• Research into Amyloidogenesis: Facilitates investigation of sequence determinants of light-chain amyloidogenesis using dipeptide-based models.

Methodology:

Support vector machine trained on dipeptide composition features derived from a dataset of 742 amyloidogenic and 712 non-amyloidogenic antibody light chains, with performance evaluated by area under the ROC curve (AUC = 0.9651).