Please refer to the git repository (https://github.com/fhvermei/chemprop_solvation) for detailed installation guide, sample code, citation, and license information
Please cite our publication: https://doi.org/10.1021/acs.jcim.1c01103
Short installation guidelines <\b>
conda install –c fhvermei -c conda-forge chemprop_solvation
pip install git+https://github.com/bp-kelley/descriptastorus
Run prediction for (1) solvation free energies Gsolv, (2) solvation enthalpies Hsolv, and (3) Abraham solute parameters:
from chemprop_solvation.solvation_estimator import load_DirectML_Gsolv_estimator, load_DirectML_Hsolv_estimator, load_SoluteML_estimator
Both Gsolv and Hsolv are in kcal/mol
smiles = [['CC1=NC=CC=C1', 'CCCCCCCC'], ... ] # solvent, solute
average_prediction, epistemic_uncertainty, valid_indices = load_DirectML_Gsolv_estimator()(smiles)
average_prediction, epistemic_uncertainty, valid_indices = load_DirectML_Hsolv_estimator()(smiles)
SoluteML predicts the Abraham solute parameters in the order of E, S, A, B, L.
smiles = [['CCCCCCCC'], ... ] # solute
average_prediction, epistemic_uncertainty, valid_indices = load_SoluteML_estimator()(smiles)
