Skip to content

LigEGFR_conda.md

Latest commit

 

History

History
156 lines (121 loc) · 4.05 KB

LigEGFR_conda.md

File metadata and controls

156 lines (121 loc) · 4.05 KB
Raw

LigEGFR prediciton based-on Anaconda installation

This method supports for Linux and macOS operating systems.

Prerequisites:

Linux with GPU

  1. Download Miniconda3-py37_4.8.3
wget -c https://repo.anaconda.com/miniconda/Miniconda3-py37_4.8.3-Linux-x86_64.sh -O Miniconda3-py37_4.8.3.sh

or

curl https://repo.anaconda.com/miniconda/Miniconda3-py37_4.8.3-Linux-x86_64.sh -o Miniconda3-py37_4.8.3.sh
  1. Install Miniconda 3 and package dependencies
chmod +x Miniconda3-py37_4.8.3.sh
./Miniconda3-py37_4.8.3.sh -b -f -p ./conda
source ./conda/bin/activate

pip install scikit-learn==0.22
conda install -y -c rdkit rdkit
pip install joblib pandas tensorflow==1.15 pillow deepchem
conda install -y -c pytorch pytorch==1.5.1 
conda install -y -c conda-forge jpype1=0.7.5
conda install -y -c openbabel openbabel=2.4.1
  1. Download a LigEGFR_source.tar.gz file from https://zenodo.org/record/4362561 and then decompress file
tar -xzvf LigEGFR_source.tar.gz

Linux without GPU (CPU only)

  1. Download Miniconda3-py37_4.8.3
wget -c https://repo.anaconda.com/miniconda/Miniconda3-py37_4.8.3-Linux-x86_64.sh -O Miniconda3-py37_4.8.3.sh

or

curl https://repo.anaconda.com/miniconda/Miniconda3-py37_4.8.3-Linux-x86_64.sh -o Miniconda3-py37_4.8.3.sh
  1. Install Miniconda 3 and package dependencies
chmod +x Miniconda3-py37_4.8.3.sh
./Miniconda3-py37_4.8.3.sh -b -f -p ./conda
source ./conda/bin/activate

pip install scikit-learn==0.22
conda install -y -c rdkit rdkit
pip install joblib pandas tensorflow==1.15 pillow deepchem
conda install -y -c pytorch pytorch==1.5.1 cpuonly
conda install -y -c conda-forge jpype1=0.7.5
conda install -y -c openbabel openbabel=2.4.1
  1. Download a LigEGFR_source.tar.gz file from https://zenodo.org/record/4362561 and then decompress file
tar -xzvf LigEGFR_source.tar.gz

macOS

  1. Download Miniconda3-py37_4.8.3
wget -c https://repo.anaconda.com/miniconda/Miniconda3-py37_4.8.3-MacOSX-x86_64.sh -O Miniconda3-py37_4.8.3.sh

or

curl https://repo.anaconda.com/miniconda/Miniconda3-py37_4.8.3-MacOSX-x86_64.sh -o Miniconda3-py37_4.8.3.sh
  1. Install Miniconda 3 and package dependencies
chmod +x Miniconda3-py37_4.8.3.sh
./Miniconda3-py37_4.8.3.sh -b -f -p ./conda
source ./conda/bin/activate

pip install scikit-learn==0.22
conda install -y -c rdkit rdkit
pip install joblib pandas tensorflow==1.15 pillow deepchem
conda install -y -c pytorch pytorch==1.5.1 
conda install -y -c conda-forge jpype1=0.7.5
conda install -y -c openbabel openbabel=2.4.1
  1. Download a LigEGFR_source.tar.gz file from https://zenodo.org/record/4362561 and then decompress file
tar -xzvf LigEGFR_source.tar.gz

How to run:

Basically, all files and folders in LigEGFR_source folder should be under the same folder that contains conda directory. The organization of folders and files is shown below.

A valid organization of a directory:

├── cdk-2.3.jar
├── conda
│   ├── *.*
├── examples
│   ├── structure*.sdf
├── models
│   ├── __init__.py
│   ├── applicabilityDomain.py
│   ├── cleanSmiles.py
│   ├── csgen.py
│   ├── fingerprint.py
│   ├── knnpickle_file
│   ├── longlist.py
│   ├── model.py
│   ├── model256.model
│   ├── mydataset.py
│   └── tobject.py
└── run_predict.py

To getting started, the conda environment requires to activate by

source conda/bin/activate

For prediction, please run run_predict.py and subsequent qeury.sdf, this file consists of multi-compounds in sdf format.

python3 run_predict.py query.sdf

When the script is computed to finish, returning a LigEGFR_output.csv file for prediction results.

Examples:

Run structure10.sdf file, this file contains 10 small molecules.

source conda/bin/activate
python3 run_predict.py examples/structure10.sdf

Run structure250.sdf file, this file contains 250 small molecules.

source conda/bin/activate
python3 run_predict.py examples/structure250.sdf