Advanced Antibody Co-Design: Antibody Diffusion Properties arrives on Vecura
This update enables protein engineers and computational biologists to co-design antibody CDR loops with improved developability properties directly through an integrated workflow on Vecura, bypassing the need for complex local infrastructure.
What is Antibody Diffusion Properties?
Antibody Diffusion Properties is an advanced, score-based generative model designed for the co-design of antibody CDR sequences and structures. By extending the established DiffAb framework, it integrates explicit guidance mechanisms to steer the generative process toward therapeutically favorable developability characteristics.
It helps users design antibodies that are not only binding-competent in silico but also optimized for solubility and aggregation propensity. It is especially useful for researchers looking to bridge the gap between initial binding discovery and downstream therapeutic developability requirements.
What can users do with Antibody Diffusion Properties on Vecura?
With Antibody Diffusion Properties on Vecura, users can:
- Perform de novo co-design of all six CDR loops simultaneously while keeping the antigen structure constant.
- Apply property-aware priors to bias initial noise toward desired hydropathy profiles.
- Utilize sampling-by-property guidance to iteratively optimize for improved binding affinity (ddG) and hydropathy during the diffusion process.
- Employ an "optimize" mode for conservative refinement of native CDR structures, allowing for localized sequence improvements.
What the output means
The output provides redesigned antibody-antigen complex PDB structures and a metadata summary of the design process.
This output should be used to support scientific decision making. It does not replace experimental validation.
Why this matters
The transition from theoretical antibody binding to a viable therapeutic candidate is often hindered by "developability" failures—issues such as poor solubility or high aggregation propensity that appear only after successful binding is achieved. Traditional models often generate binders that fail these rigorous downstream filters, requiring iterative, manual, and time-consuming adjustments.
By injecting property awareness directly into the diffusion process, Antibody Diffusion Properties shifts the paradigm from post-hoc filtering to proactive design. This allows researchers to explore the sequence-structure space with therapeutic constraints in mind from the beginning, significantly increasing the probability of discovering candidates that are both effective and developable.
- Developed by: Research group associated with the referenced PRX Life 2024 paper.
- Source: Official GitHub repository (https://github.com/amelvim/antibody-diffusion-properties)
- Reference: PRX Life 2024
在 Vecura 上试用 Antibody Diffusion Properties
打开模型工作区,用您自己的输入开始评估