First Winners Emerge in the “Race” to Open-Source AlphaFold 3

Author(s): LucianoSphere (Luciano Abriata, PhD)

Originally published on Towards AI.

This member-only story is on us. Upgrade to access all of Medium.

“A race of AI models to predict protein structures” as imagined by Dall-E-3 via ChatGPT.

The development and release of AlphaFold by Deepmind, more so of AlphaFold 2 which is not just a tweaked version of AlphaFold 1 but entailed a whole re-thinking of the problem using totally new computer science maths and concepts, revolutionized the field of protein structure prediction. And thus, biology. Moreover, besides being revolutionary in itself, Deepmind making AF2 open source allowing various breakthroughs in computational molecular sciences and fostering the development, by academics and private companies, of a whole new generation of AI models inspired by its architecture, as I have covered in this publication and in other places on Medium.

However, its latest star model, AlphaFold 3, which is “molecularly multimodal” (by which I mean it can understand not just proteins but all kinds of biologically relevant molecules) is totally closed. Google and Deepmind only provide access to it through a server, with a limited number of calls per day and only for non-commercial applications.

After heated debate about this in the scientific community, especially given how the paper reporting AF3 was published despite a lack… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI