alphafold_3.0.0.def

bootstrap: docker
from: docker.io/nvidia/cuda:12.6.0-base-ubuntu22.04

%post
    export DEBIAN_FRONTEND=noninteractive

    # Get latest package listing, install software-properties-common, git, wget,
    # compilers and libraries.
    # git is required for pyproject.toml toolchain's use of CMakeLists.txt.
    # gcc, g++, make are required for compiling hmmer and AlphaFold 3 libaries.
    # zlib is a required dependency of AlphaFold 3.
    apt update --quiet \
        && apt install --yes --quiet software-properties-common \
        && apt install --yes --quiet git wget gcc g++ make zlib1g-dev zstd

    # Get apt repository of specific Python versions. Then install Python. Tell APT
    # this isn't an interactive TTY to avoid timezone prompt when installing.
    add-apt-repository ppa:deadsnakes/ppa \
        && apt install --yes --quiet python3.11 python3-pip python3.11-venv python3.11-dev

    # Update pip to the latest version. Not necessary in Docker, but good to do when
    # this is used as a recipe for local installation since we rely on new pip
    # features for secure installs.
    python3.11 -m venv /alphafold3_venv
    PATH="/hmmer/bin:/alphafold3_venv/bin:$PATH"
    pip3 install --upgrade pip

    # Install HMMER. Do so before copying the source code, so that docker can cache
    # the image layer containing HMMER.
    mkdir /hmmer_build /hmmer ; \
        wget http://eddylab.org/software/hmmer/hmmer-3.4.tar.gz --directory-prefix /hmmer_build ; \
        (cd /hmmer_build && tar zxf hmmer-3.4.tar.gz && rm hmmer-3.4.tar.gz) ; \
        (cd /hmmer_build/hmmer-3.4 && ./configure --prefix /hmmer) ; \
        (cd /hmmer_build/hmmer-3.4 && make -j8) ; \
        (cd /hmmer_build/hmmer-3.4 && make install) ; \
        (cd /hmmer_build/hmmer-3.4/easel && make install) ; \
        rm -R /hmmer_build

    # Copy the AlphaFold 3 source code from the local machine to the container and
    # set the working directory to there.
    mkdir -p app
    cd app/
    git clone https://github.com/google-deepmind/alphafold3.git --branch v3.0.0 alphafold
    cd alphafold


    # Install the Python dependencies AlphaFold 3 needs.
    pip3 install -r dev-requirements.txt
    pip3 install --no-deps .

    # Build chemical components database (this binary was installed by pip).
    build_data

    # create directories for models and database
    mkdir -p /data
    mkdir -p /models

%environment
    # set python env
    export PATH="/hmmer/bin:/alphafold3_venv/bin:${PATH}"

    # to avoid potentially-conflicting Python packages installed in ~/.local
    export PYTHONNOUSERSITE=true

    # To work around a known XLA issue causing the compilation time to greatly
    # increase, the following environment variable setting XLA flags must be enabled
    # when running AlphaFold 3:
    export XLA_FLAGS="--xla_gpu_enable_triton_gemm=false"

    # Memory settings used for folding up to 5,120 tokens on A100 80 GB.
    export XLA_PYTHON_CLIENT_PREALLOCATE=true
    export XLA_CLIENT_MEM_FRACTION=0.95

%runscript
    exec python3 /app/alphafold/run_alphafold.py "$@"

%labels
    Author paula_sanematsu@g.harvard.edu

%help
    Singularity container to run Alphafold3, converted from the Dockerfile https://github.com/google-deepmind/alphafold3/commits/main/docker/Dockerfile