Chai One: Advancements in Drug Discovery Models and Evaluation

Chai one aims to revolutionize drug discovery by improving molecular interaction understanding for better medicines.

### Advancements in Drug Discovery: Chai One's Open-Source Model vs. AlphaFold
* 🔬  Chai One is excited to introduce its open-source model to the community, highlighting its **competitive edge** compared to AlphaFold, especially in **evolutionary scale modeling**.
* 📊  The evaluation of models in this field can be challenging due to many being available **only under non-commercial settings**, complicating benchmarking efforts.
* 🏆  In a benchmark called Pose Busters, Chai One's model achieved a score of **77%**, slightly outperforming AlphaFold's **76%**, indicating that both models are **closely matched**.
* 💰  The recent **$30 million funding** will be directed towards talent acquisition and computational resources to enhance their **drug discovery capabilities**.
* 🧪  Drug discovery is a complex process, and the team at Chai One is dedicated to developing tools that can **significantly improve this field**.
* 👥  The team, including co-founders Matt McParland, Jack Dunn, and Jack Boudreau, is passionate about the **potential impact of their work on drug discovery**.
* 🌍  Early signs of success are emerging, such as in Taiwan, where integrating **specific constraints into the model** has led to **better insights into molecular interactions**.
* 💡  By understanding and reprogramming molecular interactions, Chai One aims to contribute to the development of **improved medicines**, ultimately **benefiting humanity**.

I am so interested in Chai one and this open-source model compared to others. When we are thinking of Alpha fold, when you're thinking of evolutionary scales model, you say it's better. How on what measurements? Well, first of all, thanks so much for having me here. We're really excited to be putting out this model for the community. You're asking a great question. And you know, one of the properties about this field right now is it's actually almost challenging to evaluate these models that they compare to one another. A lot of these models are available only under non-commercial settings, which means it's hard for a company like us to actually benchmark against them. But we find one of the benchmarks, for example, a benchmark called Pose Busters. We found our models about 77% versus the 76% from Alpha three. The models are really neck and neck and we're really excited to put out a model and make it available for everyone.

including for the companies that want to engage in drug discovery. Josh Alpha Volt has the juggernaut of Google Alphabet behind it. Behind you sit some empty chairs, and we understand just a team of ten. What will that 30 million go towards? Is it about the talent? Is it about the compute? Where do you want to build? Well, we're going into really exciting times. Drug discovery is an extremely hard process. The fact that drug discovery works at all with the tools that are available today is nothing short of miraculous. So, we're really excited to kind of participate in this market and build tools that we hope will really make a difference in the process. I think the results speak for themselves. This is why we were excited, even just a couple of months of work, to put that out for the community, for folks to evaluate themselves and to hopefully find it useful. But there's so much work ahead of us, and I'm really just grateful for the

incredible team we have behind us. My co-founders, for example, Matt McParland and Jack Dunn and Jack Boudreau, among others. It's just such a great team to be working with. And I think we're just inspired by the possibilities that are ahead of us. I agree with them, and there's been a lot of hope and enthusiasm for the impact that I will have on drug discovery for a while. But you are showing improvement to success rates. When does that sort of manifest itself in the real world pushes forward the field that one day a consumer or a patient gets access to this work? We're in the really early innings of drug discovery. We're seeing those first signs of life. In Taiwan, for example, we've shown that you can integrate certain constraints into the model, the kinds of things that you might derive from white lab experiments and use those in order to get significantly better improvements on understanding how molecules interact.

with one another. Those molecules are the basis of life as we know it. So we think with these kinds of new capabilities, we can use them to better understand how molecules interact and then use that understanding to reprogram them, and then hopefully come up with better medicines and just increase, you know, make humanity better for all of us.

OpenAI, Thrive Capital Back AI Drug Discovery Startup

Welcome to the Bloomberg Technology YouTube channel. Subscribe for original series and mini-docs on game-changing technologies, profiles on innovators, live conversations with experts, and breaking technology news & analysis from Bloomberg TV.

Bloomberg Technology

Chai One's Open-Source Model and Its Impact

Chai One's open-source model is designed to compete with others, like Alpha Fold, by offering similar benchmark results.

The company's $30 million funding will be used for talent and computing resources to make a significant impact in drug discovery.

The model's release aims to provide valuable tools for the community and companies involved in drug discovery, with the goal of making a difference in the challenging process.

The Impact of a Strong Team on Drug Discovery

The team behind the drug discovery effort includes co-founders Matt McParland, Jack Dunn, and Jack Boudreau, all of whom are inspired by the possibilities ahead.

There is hope and enthusiasm for the impact of their work on drug discovery, with early signs of success in integrating constraints from lab experiments to improve the understanding of molecule interactions.

The team believes they are in the early stages of making a real-world impact, with the potential to eventually provide consumers and patients access to their work.

Molecules are fundamental to life and their interactions are crucial for understanding biological processes.

New capabilities in studying these interactions aim to reprogram molecules for better medicine.

The ultimate goal is to enhance human well-being and improve life for everyone.