AlphaFold 构建报错汇总

AlphaFold 构建报错汇总

问题1(Dockerfile)

在apt-get update命令处提示:

Err:13 https://developer.download.nvidia.cn/compute/cuda/repos/ubuntu1804/x86_64 InRelease
Temporary failure resolving ‘developer.download.nvidia.cn’
Err:14 https://developer.download.nvidia.cn/compute/machine-learning/repos/ubuntu1804/x86_64 InRelease
Temporary failure resolving ‘developer.download.nvidia.cn’

并且紧接着造成连锁错误：

W: Failed to fetch https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/InRelease Temporary failure resolving ‘developer.download.nvidia.cn’
W: Failed to fetch https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64/InRelease Temporary failure resolving ‘developer.download.nvidia.cn’
W: Some index files failed to download. They have been ignored, or old ones used instead.

解决
解决方案 - 修改DNS
修改docker的守护参数
sudo nano /etc/docker/daemon.json
追加dns相关配置

"dns":["114.114.114.114","8.8.8.8","119.29.29.29","223.5.5.5"]
sudo systemctl daemon-reload
sudo systemctl restart docker

然后重新运行构建命令
docker build -f docker/Dockerfile -t alphafold .

问题解决。

问题2(Dockerfile)

docker在19.03版本之后，可以不用安装nvidia-docker，就能获得GPU的计算支持。
但是，直接在docker中使用GPU设备，出现报错如下：

docker: Error response from daemon: could not select device driver “” with capabilities: [[gpu]].

这个问题可以通过安装nvidia-container-toolkit来解决。

1、添加nvidia-docker的源

curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | \
  sudo apt-key add -
distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | \
  sudo tee /etc/apt/sources.list.d/nvidia-docker.list
sudo apt-get update

如果不添加上面的源，可能会报如下错误：
E: Unable to locate package nvidia-container-toolkit
2、安装nvidia-container-toolkit

sudo apt-get install -y nvidia-container-toolkit
3、重启docker

sudo systemctl restart docker

问题3（非Dockerfile）

使用官方的jaxlib会报错：AttributeError: module ‘jaxlib.pocketfft’ has no attribute ‘pocketfft’
或者
RuntimeError: jaxlib version 0.3.25 is newer than and incompatible with jax version 0.3.17. Please update your jax and/or jaxlib packages.

解决：
pip install --upgrade jax== 0.2.24 jaxlib==0.1.76 dm-haiku=0.0.4

问题4（非Dockerfile）

如果输出内容里出现：CUDA - Error computing forces with CUDA platform，原因在于cudatookit不对。

解决：
首先使用 nvidia-smi 查看CUDA Version，然后使用 conda install -c conda-forge cudatoolkit= CUDA Version(对应的版本号)，就可以解决问题。当然没有cuda也能跑起来，只是费时间。

问题5（非Dockerfile）

ModuleNotFoundError: No module named ‘pdbfixer’

解决
conda install -c conda-forge pdbfixer

问题6（非Dockerfile）

ModuleNotFoundError: No module named ‘mock’

解决：
pip install mock

问题7（非Dockerfile）

ValueError: Could not find path to the “jackhmmer” binary. Make sure it is installed on your system.
解决
conda install -c biocore hmmer

ValueError: Could not find path to the “kalign” binary. Make sure it is installed on your system.

解决：
conda install -c bioconda kalign3

问题8 hmmsearch failed: Parse failed (sequence file /path/pdb_seqres/pdb_seqres.txt): Line 1360234: illegal character 0

issue解决： https://github.com/deepmind/alphafold/issues/569

awk 'NR%2==0' pdb_seqres.txt | grep 0
tac pdb_seqres.txt | sed '/^CT05/,+1d' | tac > pdb_seqres_fixed.txt