Vecura
Pricing
Contact us
Vecura

Product

  • Pricing

Company

  • Contact us

Resources

  • Blogs
  • Community

Legal

  • Privacy Policy
  • Terms of Service
  • Trust Center

© 2026 NYB AI. All rights reserved.

All systems operational
Vecura
Pricing
Contact us
Back to blog

Accelerate Antibody Design: ABodyBuilder3 is Now Available on Vecura

This update enables antibody engineers and protein scientists to generate accurate 3D atomic structures of antibody variable domains directly through a guided workflow in Vecura, eliminating the need for complex local infrastructure and time-consuming manual setup.

May 12, 2026ABodyBuilder3
ABodyBuilder3
ABodyBuilder3 is now available on Vecura
vecura.com

What is ABodyBuilder3?

ABodyBuilder3 is a specialized deep-learning model designed to predict the 3D atomic structure of paired antibody variable domains (VH/VL) directly from their amino-acid sequences. Unlike general-purpose protein structure predictors, this model is trained specifically on antibody data, enabling it to deliver high-accuracy structural models in under two seconds. By eliminating the need for Multiple Sequence Alignments (MSA) or template searches, it provides a self-contained, reproducible pipeline for researchers.

It helps users rapidly generate structural insights for antibody engineering, facilitating the evaluation of CDR loop conformations. It is especially useful for early-stage therapeutic development where obtaining reliable atomic models of hypervariable regions—particularly the CDRH3 loop—is a common bottleneck.

What can users do with ABodyBuilder3 on Vecura?

With ABodyBuilder3 on Vecura, users can:

  • Predict the full 3D atomic structure of antibody Fv fragments from input sequences.
  • Retrieve PDB-formatted files for immediate integration into docking or molecular visualization workflows.
  • Assess the reliability of predictions using per-residue pLDDT confidence scores.
  • Evaluate the overall structural confidence through a mean pLDDT summary metric.

What the output means

The output provides a 3D atomic model in PDB format, alongside per-residue pLDDT confidence scores. These scores, ranging from 0 to 100, indicate the local model accuracy; values above 90 represent high confidence, while lower scores signal potentially flexible regions.

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

Why this matters

Obtaining high-quality structural models of antibodies is crucial for predicting binding interactions and engineering desired pharmacological properties. The hypervariability of the CDRH3 loop has historically made in silico modeling challenging. ABodyBuilder3 addresses this by leveraging a state-of-the-art structure module that achieves superior accuracy on these critical regions.

By offering this capability directly within the Vecura platform, researchers can bypass the technical overhead of local installation and database management, significantly accelerating the design-build-test cycle in antibody discovery.

  • Developed by: Exscientia
  • Source: Official GitHub Repository
  • Reference: Kenlay et al., Bioinformatics 40(10), 2024

Try ABodyBuilder3 on Vecura.

Open the model workspace and start evaluating it with your own inputs.

Try model

Topics

antibodystructure-predictionproteinVH/VLpLDDTdeep-learning

On this page

What is ABodyBuilder3?What can users do with ABodyBuilder3 on Vecura?What the output meansWhy this matters
Vecura

Product

  • Pricing

Company

  • Contact us

Resources

  • Blogs
  • Community

Legal

  • Privacy Policy
  • Terms of Service
  • Trust Center

© 2026 NYB AI. All rights reserved.

All systems operational