AlphaFold 2 method published in Nature with open-source code

Google DeepMind/AlphaFold

DeepMind published the full AlphaFold 2 method in Nature, describing how it predicts protein structures from amino-acid sequences at near-atomic accuracy. The source code, trained weights and inference script were released under an open-source licence, letting any lab run it.

• Predicts protein structures at near-atomic accuracy from amino-acid sequences alone.
• Source code, trained weights and inference script released under open-source licence.
• Incorporates physical and biological knowledge about protein structure.
• Uses multi-sequence alignments in deep learning algorithm design.