For each step of the workflow we have a single directory with more detailed instructions on how to reproduce each step.
- Screening of ZINC for similar molecules:
screening - Docking of the screened molecules:
docking - Simulation using MMGBSA: An example configuration of the system for can be found in
simulation_configuration - Before running Bayesian Optimization/Active learning we preprocess (e.g. deduplication) the data. See
data/process_data.ipynb
The project uses Python 3.10. All requirements for the different steps in the workflow can be found in the respective directory.
To setup Bayesian Optimization follow the instructions here.
Once setup preprocess using the data/process_data.ipynb notebook and copy the resulting Enamine10k_scores.csv, Enamine50k_scores.csv, MCL1-vina.csv, MCL1-mmgbsa.csv
to bayesian_optimization/data/processed.
To run all setups execute the run_bayesian_optimization.sh script. Once this is finished the results can be found in bayesian_optimization/runs.