GGL-ETA-Score

This code compute features for both the SYBYL-GGL and ECIF-GGL models. The folder src contains the main source code. The code get_ggl_features.py can be used to generate features for a given protein-ligand dataset.

Package Requirement

• NumPy
• SciPy
• Pandas
• BioPandas
• RDKit

Run the following command to install the necessary packages and create a conda environmet from the provided ggl-score-env.yml file.

conda env create -f ggl-score-env.yml

Simple Example

Assume we want to genrate the features for the PDBbind v2016 general set for both SYBYL GGL and ECIF GGL with exponential kernel type and parameters $\kappa=2.5$ and $\tau=1.5$ which is the index 84 of the kernels.csv file in the utils folder. Assume also the structures of the dataset are in the directory ../PDBbind_v2016_general_Set and we wish to save the features in the directory ../Features.

# Generate the SYBYL GGL features
python get_ggl_features.py -k 84 -c 12.0 -m 'SYBYL' -f '../csv_data_file/PDBbindv2016_GeneralSet.csv' -dd '../PDBbind_v2016_general_set' -fd '../Features'

# Generate the ECIF GGL features
python get_ggl_features.py -k 84 -c 12.0 -m 'ECIF' -f '../csv_data_file/PDBbindv2016_GeneralSet.csv' -dd '../PDBbind_v2016_general_set' -fd '../Features'