SPIRED-Fitness

An end-to-end framework for the prediction of protein structure and fitness from single sequence

Install software on Linux (Conda)

1. Download SPIRED-Fitness

git clone https://github.com/Gonglab-THU/SPIRED-Fitness.git
cd SPIRED-Fitness

2. Install Anaconda / Miniconda software

3. Follow the steps below to install the Python package or run the command conda env create -f environment/spired_fitness.yml

conda create -n spired_fitness python=3.11
conda activate spired_fitness

conda install pytorch cpuonly -c pytorch
pip install click==8.1.7
pip install einops==0.7.0
pip install pandas==2.1.4
pip install biopython==1.82

4. Follow the steps below to install GDFold2 or run the command conda env create -f environment/gdfold2.yml

:exclamation: install PyRosetta at PyRosetta LICENSE

conda create -n gdfold2 python=3.11
conda activate gdfold2

conda install pytorch pytorch-cuda=11.8 -c pytorch -c nvidia
pip install biopython==1.82

Install software on Linux (Docker)

1. Download SPIRED-Fitness

git clone https://github.com/Gonglab-THU/SPIRED-Fitness.git
cd SPIRED-Fitness

2. Build Docker image from Dockerfile

docker build -t spired -f ./environment/Dockerfile .

3. Start Docker container from the image for the first time

Note: Please ensure that the container has a minimum of 50 GB of memory to run the models. Also, remember to adjust the Docker memory setting accordingly.

docker run -it --name=spired --memory=50g spired /bin/bash

Usage

• You should download the model parameters and move it into the model folder. If you install from docker, then the model parameters have been pre-downloaded.

# run SPIRED
bash run_spired.sh -i example_spired/test.fasta -o example_spired

# run SPIRED-Fitness
bash run_spired_fitness.sh -i example_fitness/test.fasta -o example_fitness

# run SPIRED-Stab
bash run_spired_stab.sh -i example_stab/test.fasta -o example_stab

Reference

An end-to-end framework for the prediction of protein structure and fitness from single sequence