Skip to content
/ MULTICOM3 Public

The software system of improving AlphaFold2- and AlphaFold-Multimer-based protein tertiary & quaternary structure prediction. It was developed by the Bioinformatics and Machine Learning Lab at the University of Missouri. It was blindly tested in CASP15 and ranked among the best server predictors in 2022. It improves AlphaFold's accuracy by 5-10%.

28 stars 8 forks Branches Tags Activity
Star
Notifications You must be signed in to change notification settings

BioinfoMachineLearning/MULTICOM3

BranchesTags

Repository files navigation

MULTICOM3

MULTICOM3 is an add-on package to improve AlphaFold2- and AlphaFold-Multimer-based prediction of protein tertiary and quaternary structures by diverse multiple sequence alignment sampling, template identification, structural prediction evaluation and structural prediction refinement. It can improve AlphaFold2-based tertiary structure prediction by 8-10% and AlphaFold-Multimer-based quaternary structure prediction by 5-8%. In CASP15, MULTICOM3 used AlphaFold v2.2.0 as the engine to generate structural predictions. In this release, it is adjusted to run on top of AlphaFold v2.3.2 (https://github.com/deepmind/alphafold/releases/tag/v2.3.2) to leverage the latest improvement on AlphaFold2. You can install MULTICOM3 on top of your AlphaFold2 and AlphaFold-Multimer to improve both the tertiary structure prediction of monomers and the quaternary structure prediction of multimers.

Overall workflow for the MULTICOM Protein tertiary structure prediction system

CASP15_TS pipeline

Overall workflow for the MULTICOM Protein complex structure prediction system

CASP15_QS pipeline

Download MULTICOM3 package

git clone --recursive https://github.com/BioinfoMachineLearning/MULTICOM3

Note: ideally, a computer with a NVIDIA V100 or better GPU, 40 or more GB GPU memory, and 85 GB or more RAM is needed to run the system.

Installation (Docker version, modified from alphafold v2.3.2)

  1. Install Docker.

  2. Check that AlphaFold will be able to use a GPU by running:

    docker run --rm --gpus all nvidia/cuda:11.0-base nvidia-smi

    The output of this command should show a list of your GPUs. If it doesn't, check if you followed all steps correctly when setting up the NVIDIA Container Toolkit or take a look at the following NVIDIA Docker issue.

    If you wish to run AlphaFold using Singularity (a common containerization platform on HPC systems) we recommend using some of the third-party Singularity setups as linked in google-deepmind/alphafold#10 or google-deepmind/alphafold#24.

  3. Build the Docker image:

    docker build -f docker/Dockerfile -t multicom3 .

    If you encounter the following error:

    W: GPG error: https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64 InRelease: The following signatures couldn't be verified because the public key is not available: NO_PUBKEY A4B469963BF863CC
E: The repository 'https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64 InRelease' is not signed.

    use the workaround described in google-deepmind/alphafold#463 (comment).

  4. Install the run_docker.py dependencies. Note: You may optionally wish to create a Python Virtual Environment to prevent conflicts with your system's Python environment.

    conda create -n docker python=3.8
conda activate docker
pip3 install -r docker/requirements.txt
conda install -c conda-forge -c bioconda mmseqs2=14.7e284 -y

  5. Make sure that the output directory exists (the default is /tmp/multicom3) and that you have sufficient permissions to write into it.

  6. Download Genetic databases in AlphaFold2/AlphaFold-Multimer

    bash $MULTICOM3_INSTALL_DIR/tools/alphafold-v2.3.2/scripts/download_all_data.sh <YOUR_ALPHAFOLD_DB_DIR>

    Note: The download directory <YOUR_ALPHAFOLD_DB_DIR> should not be a subdirectory in the MULTICOM3 repository directory. If it is, the Docker build will be slow as the large databases will be copied during the image creation.

  7. Download additional genetic databases and tools in MULTICOM3

    python download_database_and_tools.py --multicom3db_dir <YOUR_MULTICOM3_DB_DIR>

    Note: The download directory <YOUR_MULTICOM3_DB_DIR> should not be a subdirectory in the MULTICOM3 repository directory. If it is, the Docker build will be slow as the large databases will be copied during the image creation.

Installation (non Docker version)

Install AlphaFold/AlphaFold-Multimer and other required third-party packages (modified from alphafold_non_docker)

Install miniconda

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh && bash Miniconda3-latest-Linux-x86_64.sh

Create a new conda environment and update

conda create --name multicom3 python==3.8
conda update -n base conda

Activate conda environment

conda activate multicom3

Install dependencies

  • Change cudatoolkit==11.2.2 version if it is not supported in your system
conda install -y -c conda-forge openmm==7.5.1 cudatoolkit==11.2.2 pdbfixer
conda install -y -c bioconda hmmer hhsuite==3.3.0 kalign2
  • Change jaxlib==0.3.25+cuda11.cudnn805 version if this is not supported in your system
pip install absl-py==1.0.0 biopython==1.79 chex==0.0.7 dm-haiku==0.0.9 dm-tree==0.1.6 immutabledict==2.0.0 jax==0.3.25 ml-collections==0.1.0 numpy==1.21.6 pandas==1.3.4 protobuf==3.20.1 scipy==1.7.0 tensorflow-cpu==2.9.0

pip install --upgrade --no-cache-dir jax==0.3.25 jaxlib==0.3.25+cuda11.cudnn805 -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html

Download chemical properties to the common folder

# Replace $MULTICOM3_INSTALL_DIR with your MULTICOM3 installation directory

wget -q -P $MULTICOM3_INSTALL_DIR/tools/alphafold-v2.3.2/alphafold/common/ https://git.scicore.unibas.ch/schwede/openstructure/-/raw/7102c63615b64735c4941278d92b554ec94415f8/modules/mol/alg/src/stereo_chemical_props.txt

Apply OpenMM patch

# Replace $MULTICOM3_INSTALL_DIR with your MULTICOM3 installation directory

cd ~/anaconda3/envs/multicom3/lib/python3.8/site-packages/ && patch -p0 < $MULTICOM3_INSTALL_DIR/tools/alphafold-v2.3.2/docker/openmm.patch

# or

cd ~/miniconda3/envs/multicom3/lib/python3.8/site-packages/ && patch -p0 < $MULTICOM3_INSTALL_DIR/tools/alphafold-v2.3.2/docker/openmm.patch

Install other required third-party packages

conda install tqdm
conda install -c conda-forge -c bioconda mmseqs2=14.7e284 -y

Download Genetic databases in AlphaFold2/AlphaFold-Multimer

# Replace $MULTICOM3_INSTALL_DIR with your MULTICOM3 installation directory

bash $MULTICOM3_INSTALL_DIR/tools/alphafold-v2.3.2/scripts/download_all_data.sh <YOUR_ALPHAFOLD_DB_DIR>

Install the MULTICOM3 addon system and its databases

# Note: here the parameters should be the absolute paths
python download_database_and_tools.py --multicom3db_dir <YOUR_MULTICOM3_DB_DIR>

# Configure the MULTICOM3 system
# Replace $MULTICOM3_INSTALL_DIR with your MULTICOM3 installation directory
# Replace $YOUR_ALPHAFOLD_DB_DIR with your downloaded AlphaFold databases directory

python configure.py --envdir ~/miniconda3/envs/multicom3 --multicom3db_dir <YOUR_MULTICOM3_DB_DIR> --afdb_dir <YOUR_ALPHAFOLD_DB_DIR>

# e.g, 
# python download_database_and_tools.py \
# --multicom3db_dir /home/multicom3/tools/multicom3_db

# python configure.py \
# --multicom3db_dir /home/multicom3/tools/multicom3_db \
# --afdb_dir /home/multicom3/tools/alphafold_databases/

The configure.py python script will

  • Copy the alphafold_addon scripts
  • Create the configuration file (bin/db_option) for running the system

Genetic databases used by MULTICOM3

Assume the following databases have been installed as a part of the AlphaFold2/AlphaFold-Multimer installation

Additional databases will be installed for the MULTICOM system by setup.py:

Important parameter values in db_option

# AlphaFold2 parameters
monomer_num_ensemble = 1
monomer_num_recycle = 3
num_monomer_predictions_per_model = 1
monomer_model_preset = monomer

# AlphaFold-Multimer parameters
multimer_num_ensemble = 1
multimer_num_recycle = 3
num_multimer_predictions_per_model = 5
multimer_model_preset = multimer

# Common parameters
alphafold_benchmark = True
use_gpu_relax = True
models_to_relax = ALL # ALL, BEST, NONE
max_template_date = 2024-06-01

Please refer to AlphaFold2 to understand the meaning of the parameters. The parameter values stored in bin/db_option file are applied to all the AlphaFold2/AlphaFold-Multimer variants in the MULTICOM3 system to generate predictions.

For Docker version installation, you can change the default parameter values in docker/db_option.

For non Docker version of the installation, the default bin/db_option file is created automatically by configure.py during the installation. The default parameter values above can be changed if needed.

Before running the system for non Docker version

Activate your python environment and add the MULTICOM3 system path to PYTHONPATH

conda activate multicom3

# Replace $MULTICOM3_INSTALL_DIR with your MULTICOM3 installation directory (absolute path)
export PYTHONPATH=$MULTICOM3_INSTALL_DIR

# e.g, 
# conda activate MULTICOM3
# export PYTHONPATH=/home/multicom3/MULTICOM3

Now MULTICOM3 is ready for you to make predictions.

Running the monomer/tertiary structure prediction pipeline

Say we have a monomer with the sequence <SEQUENCE>. The input sequence file should be in the FASTA format as follows:

>sequence_name
<SEQUENCE>

Note: It is recommended that the name of the sequence file in FASTA format should be the same as the sequence name.

Then run the following command:

# Please provide absolute path for the input parameters
# docker version
python3 docker/run_docker.py \
    --mode=monomer \
    --option_file=docker/db_option \
    --fasta_path=$YOUR_FASTA \
    --run_img=False \
    --af_db_dir=$YOUR_ALPHAFOLD_DB_DIR \ 
    --multicom3_db_dir=$YOUR_MULTICOM3_DB_DIR \
    --output_dir=$OUTDIR

# non docker version
python bin/monomer.py \
    --option_file=bin/db_option \
    --fasta_path=$YOUR_FASTA \
    --run_img=False \
    --output_dir=$OUTDIR

option_file is a file in the MULTICOM package to store some key parameter values for AlphaFold2 and AlphaFold-Multimer. fasta_path is the full path of the file storing the input protein sequence(s) in the FASTA format. output_dir specifies where the prediction results are stored. Please be aware that we have included a parameter (--run_img) that allows you to turn off the usage of the IMG database for faster prediction (--run_img=False). In the case of --run_img=True, the program will pause at the monomer prediction generation stage to wait for the IMG alignment to be created. Generating alignments from IMG may take a much longer time, potentially several days, because the database is very large. So run_img is set to false by default. It is advised that run_img is set to true only if other alignments cannot yield good results.

Output

$OUTPUT_DIR/                                   # Your output directory
    N1_monomer_alignments_generation/          # Working directory for generating monomer MSAs
    N1_monomer_alignments_generation_img/      # Working directory for generating IMG MSA
        # Note: the img.running file may use many disk space
    N2_monomer_template_search/                # Working directory for searching monomer templates
    N3_monomer_structure_generation/           # Working directory for generating monomer structural predictions
    N4_monomer_structure_evaluation/           # Working directory for evaluating the monomer structural predictions
        - alphafold_ranking.csv    # AlphaFold2 pLDDT ranking
        - pairwise_ranking.tm      # Pairwise (APOLLO) ranking
        - pairwise_af_avg.ranking  # Average ranking of the two
    N5_monomer_structure_refinement_avg/       # Working directory for monomer structure refinement
    N5_monomer_structure_refinement_avg_final/ # Output directory for the refined monomer predictions
        - final_ranking.csv        # AlphaFold2 pLDDT ranking of the original and refined predictions

  • The predictions and ranking files are saved in the N4_monomer_structure_evaluation folder. You can check the AlphaFold2 pLDDT score ranking file (alphafold_ranking.csv) to look for the structure with the highest pLDDT score. The pairwise_ranking.tm and pairwise_af_avg.ranking are the other two ranking files.

  • The refined monomer predictions are saved in N5_monomer_structure_refinement_avg_final.

Running the multimer/quaternary structure prediction pipeline

Folding a homo-multimer

Say we have a homomer with 4 copies of the same sequence <SEQUENCE>. The input file should be in the format as follows:

>sequence_1
<SEQUENCE>
>sequence_2
<SEQUENCE>
>sequence_3
<SEQUENCE>
>sequence_4
<SEQUENCE>

Then run the following command:

# Please provide absolute path for the input parameters
# docker version
python3 docker/run_docker.py \
    --mode=homomer \
    --option_file=docker/db_option \
    --fasta_path=$YOUR_FASTA \
    --run_img=False \
    --af_db_dir=$YOUR_ALPHAFOLD_DB_DIR \ 
    --multicom3_db_dir=$YOUR_MULTICOM3_DB_DIR \
    --output_dir=$OUTDIR

# non docker version
python bin/homomer.py \
    --option_file=bin/db_option \
    --fasta_path=$YOUR_FASTA \
    --run_img=False \
    --output_dir=$OUTDIR

Folding a hetero-multimer

Say we have an A2B3 heteromer, i.e. with 2 copies of <SEQUENCE A> and 3 copies of <SEQUENCE B>. The input file should be in the format as follows (the same sequences should be grouped together):

>sequence_1
<SEQUENCE A>
>sequence_2
<SEQUENCE A>
>sequence_3
<SEQUENCE B>
>sequence_4
<SEQUENCE B>
>sequence_5
<SEQUENCE B>

Then run the following command:

# Please provide absolute path for the input parameters
# docker version
python3 docker/run_docker.py \
    --mode=heteromer \
    --option_file=docker/db_option \
    --fasta_path=$YOUR_FASTA \
    --run_img=False \
    --af_db_dir=$YOUR_ALPHAFOLD_DB_DIR \ 
    --multicom3_db_dir=$YOUR_MULTICOM3_DB_DIR \
    --output_dir=$OUTDIR

# non docker version
python bin/heteromer.py \
    --option_file=bin/db_option \
    --fasta_path=$YOUR_FASTA \
    --run_img=False \
    --output_dir=$OUTDIR

Output

$OUTPUT_DIR/                                   # Your output directory
    N1_monomer_alignments_generation/          # Working directory for generating monomer MSAs
        - Subunit A
        - Subunit B
        - ...
    N1_monomer_alignments_generation_img/      # Working directory for generating IMG MSA
        - Subunit A
        - Subunit B
        - ...
    N2_monomer_template_search/                # Working directory for searching monomer templates
        - Subunit A
        - Subunit B
        - ...
    N3_monomer_structure_generation/           # Working directory for generating monomer structural predictions
        - Subunit A
        - Subunit B
        - ...
    N4_monomer_alignments_concatenation/       # Working directory for concatenating the monomer MSAs
    N5_monomer_templates_search/               # Working directory for concatenating the monomer templates
    N6_multimer_structure_generation/          # Working directory for generating multimer structural predictions
    N7_monomer_structure_evaluation            # Working directory for evaluating monomer structural predictions
        - Subunit A
            # Rankings for all the predictions
            - alphafold_ranking.csv            # AlphaFold2 pLDDT ranking 
            - pairwise_ranking.tm              # Pairwise (APOLLO) ranking
            - pairwise_af_avg.ranking          # Average ranking of the two 

            # Rankings for the predictions generated by monomer structure prediction
            - alphafold_ranking_monomer.csv    # AlphaFold2 pLDDT ranking 
            - pairwise_af_avg_monomer.ranking  # Average ranking 

            # Rankings for the predictions extracted from multimer predictions
            - alphafold_ranking_multimer.csv   # AlphaFold2 pLDDT ranking 
            - pairwise_af_avg_multimer.ranking # Average ranking 

        - Subunit B
        - ...
    N8_multimer_structure_evaluation           # Working directory for evaluating multimer structural predictions
        - alphafold_ranking.csv                # AlphaFold2 pLDDT ranking
        - multieva.csv                         # Pairwise ranking using MMalign
        - pairwise_af_avg.ranking              # Average ranking of the two
    N9_multimer_structure_refinement           # Working directory for refining multimer structural predictions
    N9_multimer_structure_refinement_final     # Output directory for the refined multimer predictions

  • The predictions and ranking files are saved in N8_multimer_structure_evaluation, similarly, you can check the AlphaFold-Multimer confidence score ranking file (alphafold_ranking.csv) to look for the structure with the highest predicted confidence score generated by AlphaFold-Multimer. The multieva.csv and pairwise_af_avg.ranking are the other two ranking files.

  • The refined multimer predictions are saved in N9_multimer_structure_refinement_final.

  • The monomer structures and ranking files are saved in N7_monomer_structure_evaluation if you want to check the predictions and rankings for the monomer structures.

Some CASP15 Prediction Examples (MULTICOM versus the Standard AlphaFold method: NIBS-AF2-multimer)

CASP15 pipeline CASP15 pipeline

Citing this work

If you use this package for tertiary or quaternary structure prediction, please cite:

Tertiary (monomer) structure prediction

Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., ... & Hassabis, D. (2021). Highly accurate protein structure prediction with AlphaFold. Nature, 596(7873), 583-589.

Liu, J., Guo, Z., Wu, T., Roy, R. S., Chen, C., & Cheng, J. (2023). Improving AlphaFold2-based protein tertiary structure prediction with MULTICOM in CASP15. Communications Chemistry, 6(1), 188. (https://www.nature.com/articles/s42004-023-00991-6)

Quaternary (multimer) structure prediction

Evans, R., O’Neill, M., Pritzel, A., Antropova, N., Senior, A., Green, T., ... & Hassabis, D. (2021). Protein complex prediction with AlphaFold-Multimer. BioRxiv, 2021-10.

Liu, J., Guo, Z., Wu, T., Roy, R. S., Quadir, F., Chen, C., & Cheng, J. (2023). Enhancing alphafold-multimer-based protein complex structure prediction with MULTICOM in CASP15. Communications Biology, 6(1), 1140. (https://www.nature.com/articles/s42003-023-05525-3)

About

The software system of improving AlphaFold2- and AlphaFold-Multimer-based protein tertiary & quaternary structure prediction. It was developed by the Bioinformatics and Machine Learning Lab at the University of Missouri. It was blindly tested in CASP15 and ranked among the best server predictors in 2022. It improves AlphaFold's accuracy by 5-10%.

Resources

Readme
Activity
Custom properties

Stars

28 stars

Watchers

3 watching

Forks

8 forks
Report repository

Releases 6

v2.1.1 Latest
Oct 22, 2023
+ 5 releases

Packages

No packages published

Contributors 2

  •  
  •  

Languages