Accelerate Antibody Discovery with IgGM on Vecura

What is IgGM?

IgGM (Immunoglobulin Generative Model) is a generative foundation model developed by Tencent AI for Science that addresses the core challenges in therapeutic antibody and nanobody discovery. By employing an SO(3) diffusion process over backbone orientations coupled with a powerful protein language model, it enables the joint sampling of CDR sequences and 3D backbone structures conditioned on specific antigen targets. This innovative approach allows researchers to generate candidates that are complementary in both sequence composition and 3D geometry.

What can users do with IgGM on Vecura?

With IgGM on Vecura, users can:

• Perform CDR Co-design: Generate novel CDR sequences and backbone structures simultaneously from scratch, conditioned on a target antigen.

• Execute Inverse Design: Perform sequence-only design from a known complex backbone to optimize binding properties.

• Humanize Antibodies: Redesign framework regions to improve the clinical profile of existing candidates.

• Conduct Affinity Maturation: Generate mutant libraries from a wild-type seed to systematically enhance binding affinity.

• Automate Epitope Calculation: Identify key interface residues from existing complex structures to focus design efforts.

What the output means

The output provides designed antibody/nanobody-antigen complex PDB files (including backbone coordinates) and FASTA files containing the optimized sequences. Additionally, the model can output computed epitope residue indices.

This output should be used to support scientific decision making. It does not replace experimental validation.

Why this matters

The discovery of therapeutic antibodies is a high-stakes, time-consuming process that requires precise matching between the antibody paratope and the antigen epitope. Traditional methods often struggle to balance structural geometry with sequence diversity, leading to low success rates in experimental binding assays.

IgGM provides a streamlined, AI-driven framework that bridges this gap, allowing for rapid generation and virtual screening of high-quality candidates. By integrating structure and sequence co-design into a single workflow, it significantly accelerates the early stages of drug discovery, enabling researchers to explore a much broader landscape of potential binders more efficiently.

• Developed by: Tencent AI for Science

• Source: IgGM GitHub Repository

• Reference: ICLR 2025 Paper and 2025 bioRxiv Preprint