Uni-Fold: Training your own deep protein-folding models

This package provides an implementation of a trainable, Transformer-based deep protein folding model. We modified the open-source code of DeepMind AlphaFold v2.0 and provided code to train the model from scratch. See the reference and the repository of DeepMind AlphaFold v2.0. To train your own Uni-Fold models, please follow the steps below:

1. Install the environment.

Run the following code to install the dependencies of Uni-Fold:

  conda create -n unifold python=3.8.10 -y
  conda activate unifold
  ./install_dependencies.sh

Uni-Fold has been tested for Python 3.8.10, CUDA 11.1 and OpenMPI 4.1.1. We recommend using Conda >= 4.10 to install the environment: using

To finish reading, please visit source site