This is the reference implementation of recurrent geometric network-2 (RGN2), described in the paper Single-sequence protein structure prediction using a language model and deep learning.

Note: We recommend using RGN2 primarily for prediction of orphan or de novo proteins. For proteins with known homologs we suggest using AlphaFold2 or OpenFold.

Use the RGN2 Colab notebook to predict the 3D structures for orphan or de novo protein sequences.

Single-sequence protein structure prediction using a language model and deep learning, Nature Biotechnology 2022