Programming

AlphaFold 3: Google’s new AI can model DNA

AlphaFold 3: Google's new AI can model DNA

pixabay.com

The new and interesting project by Google DeepMind baptized AlphaFold 3 it promises to be a revolutionary tool in the context of genetics and beyond.

We’re talking about a AI model capable of analyzing and modeling the protein structure and of all those that are considered as the “Molecules of life“. On a practical level, AlphaFold 3 will be useful in various fields, first and foremost medicine e agricultureas well as in the development of drugs and as regards any scientific test.

As is easy to guess from the name, this is the third version of the model. Compared to the previous ones, the current one is able to go beyond proteins, interacting with DNA, RNA and also up smaller molecules (i.e to bind).

Not only that: according to what DeepMind claimed, AlphaFold 3 has a50% greater accuracy compared to its predecessor.

How AlphaFold 3 works and why it can revolutionize medicine

The AI ​​model we are talking about is based on a large library it contains molecular structures.

The researchers, in fact, fed AlphaFold 3 with a huge list of combinable molecules, with the AI ​​acting by combining them and generating the 3D model of a new structure. On a purely practical level, the operation is comparable in some ways to that of some systems that allow images to be generated with AI, a bit like what happens with Stable Diffusion.

At present, this new tool has already been used in practice. A company that creates new drugs known as Isomorphic Labsin fact, has already developed projects through this platform, with applications regarding the understanding diseases.

Google also stated how the project is shared with scientific community e political leaders for one responsible development of the model. This is very important taking into account the risks related to AI and the theoretical use in the context of the creation of biological weapons.

Leave a Reply

Your email address will not be published. Required fields are marked *