Skip to content

installation_guide.md

Latest commit

 

History

History
83 lines (57 loc) · 2 KB

installation_guide.md

File metadata and controls

83 lines (57 loc) · 2 KB

Installation guide

Prerequisites

  • OS support: Windows, Linux and OSX
  • Python version: 3.6, 3.7

Dependencies

name version
numpy -
pandas -
networkx -
paddlepaddle >=2.0.0rc0
pgl >=2.1
rdkit -
sklearn -

('-' means no specific version requirement for that package)

Instruction

Since PaddleHelix depends on the paddlepaddle of version 2.0.0rc0 or above, and rdkit cannot be installed directly using pip, we suggest using conda to create a new environment for the installation. Detailed instruction is shown below:

  1. If you do not have conda installed, please check this website to get it first:

https://docs.conda.io/projects/conda/en/latest/user-guide/install/

  1. Create a new environment with conda:
conda create -n paddlehelix python=3.7
  1. Activate the environment which is just created:
conda activate paddlehelix
  1. Install rdkit using conda:
conda install -c conda-forge rdkit

  1. Install the right version of paddlepaddle according to the device (CPU/GPU) you want to run PaddleHelix on.

    If you want to use the GPU version of paddlepaddle, run this:

    python -m pip install paddlepaddle-gpu -f https://paddlepaddle.org.cn/whl/stable.html

    Or if you want to use the CPU version of paddlepaddle, run this:

    python -m pip install paddlepaddle -i https://mirror.baidu.com/pypi/simple

    Noting that the version of paddlepaddle should be higher than 2.0. Check paddlepaddle's official document for more installation guide.

  2. Install PGL using pip:

pip install pgl
  1. Install PaddleHelix using pip:
pip install paddlehelix
  1. The installation is done!

Note

After playing, if you want to deactivate the conda environment, do this:

conda deactivate