IgPose: A Generative Data-Augmented Pipeline for Robust Immunoglobulin-Antigen Binding Prediction

TL;DR

IgPose is a novel generative data-augmentation framework that improves immunoglobulin-antigen binding prediction accuracy using synthetic decoys and advanced neural networks, aiding high-throughput antibody discovery.

Contribution

Introduces IgPose, a new pipeline combining synthetic decoys and neural networks for robust Ig-Ag binding pose prediction, addressing data scarcity and improving generalization.

Findings

Outperforms existing methods on internal and CASP-16 benchmarks.

Effective in high-throughput antibody discovery pipelines.

Utilizes synthetic decoys to enhance model robustness.

Abstract

Predicting immunoglobulin-antigen (Ig-Ag) binding remains a significant challenge due to the paucity of experimentally-resolved complexes and the limited accuracy of de novo Ig structure prediction. We introduce IgPose, a generalizable framework for Ig-Ag pose identification and scoring, built on a generative data-augmentation pipeline. To mitigate data scarcity, we constructed the Structural Immunoglobulin Decoy Database (SIDD), a comprehensive repository of high-fidelity synthetic decoys. IgPose integrates equivariant graph neural networks, ESM-2 embeddings, and gated recurrent units to synergistically capture both geometric and evolutionary features. We implemented interface-focused k-hop sampling with biologically guided pooling to enhance generalization across diverse interfaces. The framework comprises two sub-networks--IgPoseClassifier for binding pose discrimination and IgPoseScore for DockQ score estimation--and achieves robust performance on curated internal test sets and the CASP-16 benchmark compared to physics and deep learning baselines. IgPose serves as a versatile computational tool for high-throughput antibody discovery pipelines by providing accurate pose filtering and ranking. IgPose is available on GitHub (https://github.com/arontier/igpose).