Applying an Adversarial Generation Engine to Steer De Novo Miniprotein & VHH Design
- Exploring how an adversarial generation engine with continuous in silico evaluation dynamically steers de novo miniprotein design to improve multi-objective hit rates
- Discussing the use of labelled design and click chemistry–based modular assembly to enable efficient scaling from single miniproteins and VHHs to multispecific constructs
- Identifying informative labels using model heuristics to bias generative sampling towards optimized affinity, specificity, stability, and immunogenicity profiles