MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202611063370 A) filed by Indian Institute Of Technology Delhi on May 19, 2026, for System For Predicting Amino Acid Mutation Effects On Antibody-Antigen Binding Affinity And A Method Thereof.
Inventor includes Sivasubramani, Santhosh.
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: A system (104) for predicting amino acid mutation effects on antibody-antigen binding affinity comprises an input module (212) receiving an antibody variable heavy chain sequence, variable light chain sequence, antigen sequence, and mutation specification. A multi-modal feature extraction module (214) extracts a combined feature vector. A physics-constrained neural network module (216) generates a binding affinity change prediction and predicted probability using a composite loss function comprising a binding loss and physics-based regularization loss having electrostatic, hydrophobic, or region-specific constraints. A self-attention feature weighing module (218) generates a weighted feature vector and feature importance scores. A confidence stratification module (220) determines effect magnitude classification, computes a prediction confidence score, and applies filtering thereon. A prediction output module (222) generates a filtered prediction output comprising the binding affinity change prediction, predicted probability, prediction confidence score, and effect magnitude classification.
Disclaimer: Curated by HT Syndication.
