Reproduce “Bayesian Optimization of Antibodies Informed by a Generative Model of Evolving Sequences” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Alan Nawzad Amin, Nate Gruver, Yilun Kuang, Lily Li, Hunter Elliott, Calvin McCarter, Aniruddh Raghu, Peyton Greenside and 1 more

CloneBO proposes a method for efficiently optimizing antibody sequences by integrating a generative language model trained on natural clonal family evolution with Bayesian optimization. This approach yields antibodies with superior binding affinity and thermostability in wet lab experiments while requiring fewer experimental cycles compared to existing methods.

Key insights

  • A generative model trained on natural antibody clonal families can implicitly learn and represent the underlying fitness landscape and evolutionary pathways of antibodies.
  • Leveraging this biologically informed prior significantly accelerates Bayesian optimization by guiding mutations towards realistic and functional sequence space.
  • The martingale posterior approach, combined with twisted Sequential Monte Carlo, enables efficient inference and sampling of fitness functions given limited experimental measurements, which is crucial for practical application.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.