AGL-ETA-Score-Open

To compute features for the AGL-EAT-Score model, use the provided code. The main source code is located in the src folder. You can generate features for a given protein-ligand dataset using the get_agl_eat_features.py script.

Required Packages

• NumPy
• SciPy
• Pandas
• BioPandas
• RDKit

Example

To generate features for the PDBbind v2016 general set for the AGL-EAT model with an exponential kernel type and parameters $\kappa = 16.5$ and $\tau = 1.5$, located at index 112 of the kernels.csv file in the utils folder, follow these steps. Assume the dataset structures are in the ../PDBbind_v2016_general_Set directory and the features should be saved in the ../Features directory.

 # Generate the AGL extended atom type (SYBYL) features
 python get_agl_eat_features.py -k 112 -c 12 -m Adjacency -f '../csv_data_file/PDBbindv2016_GeneralSet.csv' -dd '../PDBbind_v2016_general_set' -fd '../Features'