AlphaFold2 Docker环境安装
使用Dockerfile安装AlphaFold2 的环境会遇到各种各样的问题,安装过程如下:
目录
将docker/run_docker.py中的DOWNLOAD_DIR 修改为存储数据的路径
Dockerfile文件内容解析,修改了安装HH-suit的部分
安装环境的前置工作
从github
下载源码,并完成前置工作(下载地址:deepmind/alphafold: Open source code for AlphaFold. (github.com))
docker run --rm --gpus all nvidia/cuda:11.0-base nvidia-smi
-
Install Docker.
-
Install NVIDIA Container Toolkit for GPU support.
-
Setup running Docker as a non-root user.
-
Download genetic databases (see below).
-
Download model parameters (see below).
-
Check that AlphaFold will be able to use a GPU by running:
使用提供的脚本下载需要用到的数据文件,数据目录如下:
$DOWNLOAD_DIR/ # Total: ~ 2.2 TB (download: 438 GB)
bfd/ # ~ 1.7 TB (download: 271.6 GB)
# 6 files.
mgnify/ # ~ 64 GB (download: 32.9 GB)
mgy_clusters_2018_12.fa
params/ # ~ 3.5 GB (download: 3.5 GB)
# 5 CASP14 models,
# 5 pTM models,
# LICENSE,
# = 11 files.
pdb70/ # ~ 56 GB (download: 19.5 GB)
# 9 files.
pdb_mmcif/ # ~ 206 GB (download: 46 GB)
mmcif_files/
# About 180,000 .cif files.
obsolete.dat
small_bfd/ # ~ 17 GB (download: 9.6 GB)
bfd-first_non_consensus_sequences.fasta
uniclust30/ # ~ 86 GB (download: 24.9 GB)
uniclust30_2018_08/
# 13 files.
uniref90/ # ~ 58 GB (download: 29.7 GB)
uniref90.fasta
将docker/run_docker.py
中的DOWNLOAD_DIR
修改为存储数据的路径
Dockerfile文件内容解析,修改了安装HH-suit的部分
# Copyright 2021 DeepMind Technologies Limited
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# 设置CUDA版本号
ARG CUDA=11.0
FROM nvidia/cuda:${CUDA}-cudnn8-runtime-ubuntu18.04
# FROM directive resets ARGS, so we specify again (the value is retained if
# previously set).
ARG CUDA
# Use bash to support string substitution.
SHELL ["/bin/bash", "-c"]
# 安装相关软件并清理缓存
RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
build-essential \
cmake \
cuda-command-line-tools-${CUDA/./-} \
git \
hmmer \
kalign \
tzdata \
wget \
&& rm -rf /var/lib/apt/lists/*
# 安装HH-suite 由于git访问https经常报错,改成了git
# Compile HHsuite from source.
#RUN git clone --branch v3.3.0 https://github.com/soedinglab/hh-suite.git /tmp/hh-suite \
RUN git clone --branch v3.3.0 git://github.com/soedinglab/hh-suite.git /tmp/hh-suite \
&& mkdir /tmp/hh-suite/build \
&& pushd /tmp/hh-suite/build \
&& cmake -DCMAKE_INSTALL_PREFIX=/opt/hhsuite .. \
&& make -j 4 && make install \
&& ln -s /opt/hhsuite/bin/* /usr/bin \
&& popd \
&& rm -rf /tmp/hh-suite
# 安装Miniconda
# Install Miniconda package manager.
RUN wget -q -P /tmp \
https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh \
&& bash /tmp/Miniconda3-latest-Linux-x86_64.sh -b -p /opt/conda \
&& rm /tmp/Miniconda3-latest-Linux-x86_64.sh
# 使用conda安装python包
# Install conda packages.
ENV PATH="/opt/conda/bin:$PATH"
RUN conda update -qy conda \
&& conda install -y -c conda-forge \
openmm=7.5.1 \
cudatoolkit==${CUDA_VERSION} \
pdbfixer \
pip \
python=3.7
# 标准氨基酸的键长、键角等参数
COPY . /app/alphafold
RUN wget -q -P /app/alphafold/alphafold/common/ \
https://git.scicore.unibas.ch/schwede/openstructure/-/raw/7102c63615b64735c4941278d92b554ec94415f8/modules/mol/alg/src/stereo_chemical_props.txt
# pip安装Python包
# Install pip packages.
RUN pip3 install --upgrade pip \
&& pip3 install -r /app/alphafold/requirements.txt \
&& pip3 install --upgrade jax jaxlib==0.1.69+cuda${CUDA/./} -f \
https://storage.googleapis.com/jax-releases/jax_releases.html
#对OpenMM做了一些修改
# Apply OpenMM patch.
WORKDIR /opt/conda/lib/python3.7/site-packages
RUN patch -p0 < /app/alphafold/docker/openmm.patch
#设置Docker虚拟机的入口
# We need to run `ldconfig` first to ensure GPUs are visible, due to some quirk
# with Debian. See https://github.com/NVIDIA/nvidia-docker/issues/1399 for
# details.
# ENTRYPOINT does not support easily running multiple commands, so instead we
# write a shell script to wrap them up.
WORKDIR /app/alphafold
RUN echo $'#!/bin/bash\n\
ldconfig\n\
python /app/alphafold/run_alphafold.py "$@"' > /app/run_alphafold.sh \
&& chmod +x /app/run_alphafold.sh
ENTRYPOINT ["/app/run_alphafold.sh"]
Docker环境构建步骤:
-
使用
docker build -f docker/Dockerfile -t alphafold .
创建镜像 -
使用conda安装虚拟环境
conda create -n alphafold python=3.7
-
进入conda创建的虚拟环境
conda activate alphafold
-
安装启动的依赖
pip3 install -r docker/requirements.txt
-
运行alphafold2 --fasta_paths是fasta文件的路径 max_template_data指定了使用模板的截止时间 --preset指定了运行模式 --gpu_devices选择用到的GPU
python3 docker/run_docker.py --fasta_paths=/data8t/zhengrongtao/test/P94485.fasta --max_template_data=2020-05-14 --preset=reduced_dbs --gpu_devices 1
--preset的模型
--preset=reduced_dbs
,--preset=full_dbs
or--preset=casp14
to the run command. We provide the following presets:-
reduced_dbs: This preset is optimized for speed and lower hardware requirements. It runs with a reduced version of the BFD database and with no ensembling. It requires 8 CPU cores (vCPUs), 8 GB of RAM, and 600 GB of disk space.
-
full_dbs: The model in this preset is 8 times faster than the
casp14
preset with a very minor quality drop (-0.1 average GDT drop on CASP14 domains). It runs with all genetic databases and with no ensembling. -
casp14: This preset uses the same settings as were used in CASP14. It runs with all genetic databases and with 8 ensemblings.
-
可视化
可以使用chirema进行可视化,下载地址:Download UCSF Chimera
点击open可以打开.pdb文件
或者使用chimeraX来进行预测Download UCSF ChimeraX