OpenAI has backed a six-month-old biology startup that plans to develop AI models for drug discovery.
The company, named Chai Discovery, has been founded by ex-OpenAI and Meta researchers and has reached almost $30 million in fundraising.
Prior to creating the startup, the team behind it had helped to advance the development of 20 separate drug programs. With the team members having previously worked at the likes of OpenAI, Stripe, Meta, and Google AI.
The startup’s announcement about its product was first introduced on Monday (September 9) through a detailed blog.
We’re excited to introduce @ChaiDiscovery and release Chai-1, a foundation model for molecular structure prediction that performs at the state-of-the-art across a variety of drug discovery tasks
We're releasing inference code, weights & a web interface: https://t.co/QmpbVO9Fhd pic.twitter.com/TU7xuOAaIF
— Joshua Meier (@joshim5) September 9, 2024
“We’re excited to release Chai-1, a new multi-modal foundation model for molecular structure prediction that performs at the state-of-the-art across a variety of tasks relevant to drug discovery. Chai-1 enables unified prediction of proteins, small molecules, DNA, RNA, covalent modifications, and more,” writes the Chai Discovery team.
Alongside the ChatGPT-maker, Thrive Capital, Conviction, Dimension, Neo, Lachy Groom, and Amplify Partners have all been part of the fundraising and partnership process.
This model comes at a time when Google DeepMind’s AlphaFold is one of the leaders in working on protein prediction models, with the startup now giving it some competition.
The hope is that these AI models and the funding that goes into them will help spur drug development for major diseases through the use of technology.
Drug discovery startup is “well funded” and “grateful” for strong partnerships
The model is available for all through a web interface and for commercial applications like drug discovery. One of the co-founders, Joshua Meier, took to X (formerly Twitter) to share further details.
“We tested Chai-1 across a number of benchmarks, and found that the model achieves a 77% success rate on the PoseBusters benchmark (vs. 76% by AlphaFold3).
“Unlike many existing structure prediction tools which require multiple sequence alignments (MSAs), Chai-1 can be run in single sequence mode without MSAs while preserving most of its performance.
“In addition to its frontier modeling capabilities directly from sequences, Chai-1 can be prompted with new data, e.g. restraints derived from the lab, which boost performance by double-digit percentage points.”
He also describes the company as being “well funded” and says they are “grateful for the strong partnership” of its investors.
Featured Image: AI-generated via Canva
