QwikFold provides an easy to use graphical interface for running AlphaFold protein folding prediction jobs on a local installation of AlphaFold. QwikFold accepts a sequence in FASTA format, and manages the details of running AlphaFold on behalf of the user.

Software Requirements

To start jobs from the QwikFold plugin, AlphaFold and its companion software must be installed locally. One should follow the Step-by-Step intructions bellow to properly install a local copy.

Database Downloads and Local Storage Requirements

Alphafold uses large sequence databases for genetic search. The download scripts are included in the github distribution. Total: ~ 2.2 TB (download: 438 GB)

Using QwikFold

In addition to the FASTA sequence to model, one should provide a Job ID, paths to alphafold installation, downloaded databases, and path to outputs.

Configure:

Job ID: Unique identifier for the job.
Path: Local path to AlphaFold cloned from GitHub.
Databases: Path to sequence databases downloaded for the genetic search
Output folder: Path to output results

FASTA Sequence Paste or load a sequence, with or without header. One may insert path to a file or load from disk.

Submit and Analysis (Buttons)

"Read config" - not implemented.
"Write config" - Write a script to run alphafold
"FOLD !" - Run alphafold
"Load Models" - Load all generated models
"Align Models" - Perform structural alignment of loaded models

Local AlphaFold Installation Script (Unix, bash)

#!/bin/bash
# QwikFold install instructions for AlphaFold
#
# Diego E. B. Gomes  | dgomes@auburn.edu
# Rafael C. Bernardi | rcbernardi@auburn.edu
#

# Install alphafold 2 - Ubuntu 18.04.
# https://github.com/deepmind/alphafold

# Step 0 - Modify these paths according
export ALPHAFOLD_DATASETS='/data/alphafold_dbs/'

# Step 1 - create a conda environment
conda create -n af2 python=3.8 -y

# Step 2 - activate the environment
conda activate af2

# Step 3 - Clone Alphafold from GitHub into THIS folder.
git clone https://github.com/deepmind/alphafold.git

# Step 4 - Download chemical properties to the common folder
wget -q -P alphafold/common/ https://git.scicore.unibas.ch/schwede/openstructure/-/raw/7102c63615b64735c4941278d92b554ec94415f8/modules/mol/alg/src/stereo_chemical_props.txt

# Step 5 - Go to AlphaFold folder, and set path to current DIR
cd alphafold
alphafold_path=${PWD}

# Step 6 - Install required packages
conda install -y -c nvidia cudnn==8.0.4
conda install -y -c bioconda hmmer hhsuite==3.3.0 kalign2
conda install -y -c conda-forge openmm=7.5.1 pdbfixer pip

# Step 7 - Upgrade PIP 
pip3 install --upgrade pip

# Step 8 - Install additional AlphaFold requirements using PIP
pip3 install -r ./requirements.txt
pip3 install --upgrade "jax[cuda111]" -f https://storage.googleapis.com/jax-releases/jax_releases.html

# Step 9 - Apply patch to OpenMM
python_path=$(which python)
cd $(dirname $(dirname ${python_path}))/lib/python3.8/site-packages
patch -p0 < ${alphafold_path}/docker/openmm.patch

# Step 10 (very slow) - Download AlphaFold datasets (reduced datasets ~/400Gb)
if  [ ! -d ${ALPHAFOLD_DATASETS} ] ; then 
  mkdir -p ${ALPHAFOLD_DATASETS}
fi

# Step 11 - Download Alphafold reduced datasets ( ~400Gb)
cd ${alphafold_path}
bash scripts/download_all_data.sh ${ALPHAFOLD_DATASETS} reduced_dbs

# Step 12 - Download Alphafold complete datasets ( ~2.2Tb )
cd ${alphafold_path}
bash scripts/download_all_data.sh ${ALPHAFOLD_DATASETS}