Chai Discovery

Chai-1 is available on Latch. This is a big model that predicts the properties and interactions of many types of molecules. It is competitive with AF3, RosettaFold, ESM3 on tasks that matter in discovery - eg. protein-ligand, PPI, protein-antibody. Strong technical work from the Chai Discovery team. Access at latch.bio.

Chai-1 has always been available for commercial use via our server. Today, we're also making Chai-1(r) code and weights available under an Apache 2.0 license, which permits broad commercial use. Since our launch 2.5 months ago, we've been blown away by the many ways in which the academic community has leveraged Chai-1. Today, we're making this same technology available to companies everywhere for free. You can download Chai from our GitHub repo (https://lnkd.in/ekqKHZ9b), or use it on channels like Tamarind Bio, LatchBio, Seqera, and many others.

Today we’re releasing Chai-1r, an updated version of Chai-1 with full support for restraints. Prompting Chai-1r with prior structural knowledge can boost accuracy by >2x. https://lnkd.in/euP7frVQ Custom restraints can refine predictions with knowledge obtained from scientific literature or lab assays. For example, a “contact” restraint specifies that two tokens are within a certain distance. A “pocket restraint” indicates a token should be in contact with another chain. Chai-1r excels in antibody-antigen interaction prediction with mean DockQ % of - Chai-1r (+4 epitope residues): 43.7 - Chai-1: 35.6 - Boltz (an AlphaFold3 reproduction from MIT & Genesis): 26.4 - AlphaFold2: 20.6 Restraints are available immediately both in our public codebase (https://lnkd.in/e3SGx_dj) and on our web server (https://lnkd.in/e9YfR6GV).