IgDesign: Advanced Antibody CDR Engineering Now Available on Vecura

What is IgDesign?

IgDesign is an advanced antibody CDR inverse folding model designed to generate novel complementarity-determining region (CDR) sequences conditioned on the 3D structure of an antibody-antigen complex. It effectively combines a fine-tuned structure-conditioned ProteinMPNN with a pretrained ESM2 protein language model to bridge the gap between geometric constraints and evolutionary sequence patterns. This approach allows researchers to explore the sequence space for CDR loops that are both structurally plausible and complementary to the target antigen.

It helps users perform structure-guided CDR redesign campaigns to improve binding affinity, increase developability, or generate diversity around known therapeutic antibody hits. It is especially useful for drug discovery teams seeking to optimize antibody candidates starting from a known structural template.

What can users do with IgDesign on Vecura?

With IgDesign on Vecura, users can:

• Generate novel CDR sequences by conditioning on the 3D structure of a given antibody-antigen complex.
• Optionally condition the design process on the antigen and/or the light chain to refine sequence generation.
• Utilize the independent loss scoring mechanism to rank sequence candidates by model confidence.
• Streamline the design pipeline without the need for managing complex technical infrastructure or model dependencies.

What the output means

The output provides a comprehensive table of designed CDR sequences, featuring cross-entropy loss scores that help rank the candidates. Lower loss values indicate higher model confidence in the structural fit and complementarity of the generated sequence.

This output should be used to support scientific decision making, such as prioritizing sequences for synthesis and experimental validation via SPR or ELISA. It does not replace experimental validation.

Why this matters

The design of antibody CDRs is a central challenge in therapeutic engineering, requiring a delicate balance between binding affinity, structural stability, and developability. By automating the generation of plausible CDR sequences based on atomic-level structural data, IgDesign accelerates the search for high-quality candidates that are more likely to succeed in later laboratory stages.

This innovation represents a significant step in computer-aided drug design, moving away from slow, manual protein engineering toward a predictive, data-driven approach that integrates 3D geometry with powerful language model representations.

• Developed by: Absci
• Source: Official GitHub Repository
• Reference: IgDesign: Antibody CDR inverse folding (bioRxiv)