- A total of 20 pharmokinetic propeties can be predicted using (18 classification and 2 regression models)
- The classification models predict the probability of being active (eg. toxic), where as regression models predict a numeric quantification of the pharmacokinetic or toxicity property
conda create -c conda-forge -n opendmpk rdkit==2018.09.1
conda activate opendmpk
pip install -r requirements.txt
Usage: python run_openDMPK.py [-h] [--smiles SMILES]
Example:
python run_openDMPK.py --smiles "OC(O)C(Cl)(Cl)Cl"
Results:
{'OC(O)C(Cl)(Cl)Cl': {'AmesMutagenesis': 0.0, 'AvianToxicity': 0.07,
'BBBpenetration': 0.08, 'Biodegradation': 0.37,
'CYP2c9': 0.14, 'CYP2d6': 0.09, 'CaCO2': 0.46,
'EyeCorrosion': 0.27, 'EyeIrritation': 0.83,
'HumanIntestineAbsorption': 0.97, 'HumanOralBioavailability': 0.47,
'OrganicCationTransporter2': 0.16, 'hERGG': 0.68,
'PlasmaProteinBinding': '0.3422 %',
'TetrahymenaPyriformisToxicity': '0.1885 pIGC50 (ug/L)'}}
- Build the docker image
docker build -t opendmpk .
and rundocker run --rm opendmpk
. Provide SMILES asdocker run --rm opendmpk <compound_smiles>
. - Download:
docker pull kcgovinda/opendmpk:latest
- Run the container:
docker run --rm kcgovinda/opendmpk:latest --smiles <compound_smiles>
.