MIT (Massachusetts Institute of Technology) have just created a new artificial intelligence capable of predicting the structure of proteins, which will allow the development of new therapeutics. Unlike Google DeepMind’s AlphaFold, its main competitor, Boltz-1 is accessible to everyone.
In 2018, before the arrival of Chat GPTChat GPT and other creative artificial intelligence, GoogleGoogle DeepMind has unveiled an AI for medical research. Called the “alpha fold”, it is able to predict the 3D structure. ProteinProteinImportant information for clinical research, especially for the development of new treatments.
gave Massachusetts Institute of Technologyy (MIT) has just unveiled its AI capable of predicting the structure of proteins. Called “Boltz-1”, it will have the same performance as AlphaFold3, the latest version of Google DeepMind AI. However, AlphaFold3 is not free and is not available for commercial use. Boltz-1, in contrast, is a freely licensed project.
A model is available for download on GitHub.
Just like AlphaFold3, Boltz-1 includes a propagation-style generating AI model, which better handles the amount of uncertainty in predicting complex protein structures. The researchers spent four months testing different algorithms to improve their performance and accuracy. Boltz-1 was designed from the ground up as a collaborative project. ” It is not for nothing that we call it Boltz-1 and not Boltz. This is not the end of the line. We want to get as much input from the community as possible. “, pointed out Gabriel Corso, one of the main developers.
Researchers thus want to democratize access to the latest tools in structural biology to accelerate the creation of new therapies. AmbiguousAmbiguous Discoveries in Medicine has published the source code on GitHub, where it is possible to download Boltz-1 and run it on your machine. They invite users who want to contact them on their Slack channel.
