Kinases to predict Inhibitor constant in terms of pKI (where pKi is decadic logarithm of Ki). We used the data points that were specifically represent Ki values to train and test the models.

• python==2.7.16
• pydpi==1.0
• numpy==1.16.5
• pandas==0.24.2
• tqdm==4.36.1
• scipy==1.2.2
• scikit-learn==0.20.4
• rdkit==2018.03

• Download the model file

./download_model.sh

• Set up the conda environment and activate it

conda env create -f environment.yml
conda activate kinasepki

Run ./test.sh to get the prediction for an example pair of protein sequence and a ligand SMILES. For any other inputs, run the following

python2 get_kinase_pki.py "<PROTEIN_SEQUENCE>" "<LIGAND_SMILES>"

• Build the docker image docker build -t kinasepki . and run docker run --rm kinasepki for a sample run that uses test.sh. To provide sequence and SMILES, do docker run --rm kinasepki <protein_sequence> <compound_smiles>

A docker image is also available on docker hub.

• Download the docker image docker pull sirimullalab/kinasepkipred:py2
• Run the container docker run --rm sirimullalab/kinasepkipred:py2 <protein_sequence> <compound_smiles>. To run with a built-in sample, do docker run --rm sirimullalab/kinasepkipred:py2