Teaching generative models to hallucinate
There are currently two main approaches to computational protein binder design: optimization (exemplified by BindCraft) and generative models (e.g. BoltzGen). A design campaign using either method looks very similar: you first generate a large collection of designs and then rank and filter them down to a handful for wetlab
There are currently two main approaches to computational protein binder design: optimization (exemplified by BindCraft) and generative models (e.g. BoltzGen). A design campaign using either method looks very similar: you first generate a large collection of designs and then rank and filter them down to a handful for wetlab testing. In practice the main difference is that optimization (also known as hallucination[[1]]) has high average per-design in silico quality, but is very slow; while generative models are much, much faster but average design quality is lower[[2]]. We'll show that, by…
saved by
related reading
- ~180 lines of code to win the in silico portion of the Adaptyv Nipah binding competitionblog.escalante.bio
- Minibinder design is just not that hardblog.escalante.bio
- One-shot design of functional protein binders with BindCraftnature.com
- Highly accurate protein structure prediction with AlphaFoldnature.com
- Patterns for Building LLM-based Systems & Productseugeneyan.com
- Fast and Ultra-Capable Protein Design: Advancing the Frontier Through Atomistic SE(3)-Equivariance with Genie 3biorxiv.org
- Language models of protein sequences at the scale of evolution enable accurate structure predictionbiorxiv.org
- The ML drug discovery startup trying really, really hard to not cheat (Leash Bio)owlposting.com
- [2203.07814] Competition-Level Code Generation with AlphaCodearxiv.org
- Engineering Plastic-Degrading Enzymes and PCSK9 Binders with Protein AI Toolsblog.latch.bio
- Generative ML in chemistry is bottlenecked by synthesisowlposting.com
- Diffusion models from scratchchenyang.co