paccmann_generator

Multimodal generative models for PaccMann^RL.

paccmann_generator is a package for conditional molecular design, with examples of molecule generation against gene expression profiles or protein targets. For example, see our papers:

• PaccMann^RL: De novo generation of hit-like anticancer molecules from transcriptomic data via reinforcement learning (iScience, 2021). In there, we use methods from deep reinforcement learning to bias a molecular generator to produce molecules that exhibit low IC50 against certain cell lines (code in this repo).
• Data-driven molecular design for discovery and synthesis of novel ligands: a case study on SARS-CoV-2 (Machine Learning: Science and Technology, 2021). In there, we use the same principle to bias a molecular generator to produce molecules that have high binding affinity against certain protein targets (code in this repo).

Installation

The library itself has few dependencies (see setup.py) with loose requirements. To run the example training script we provide environment files under examples/.

Create a conda environment:

conda env create -f examples/IC50/conda.yml

Activate the environment:

conda activate paccmann_generator

Install in editable mode for development:

git checkout 0.0.1  # Needed only for IC50 example (for affinity example skip this line)
pip install -e .

Example usage

In the examples/IC50 directory is a training script train_paccmann_rl.py that makes use of paccmann_generator.

(paccmann_generator) $ python examples/IC50/train_paccmann_rl.py -h
usage: train_paccmann_rl.py [-h]
                            mol_model_path omics_model_path ic50_model_path
                            smiles_language_path omics_data_path params_path
                            model_name site

PaccMann^RL training script

positional arguments:
  mol_model_path        Path to chemistry model
  omics_model_path      Path to omics model
  ic50_model_path       Path to pretrained ic50 model
  smiles_language_path  Path to SMILES language object
  omics_data_path       Omics data path to condition generation
  params_path           Model params json file directory
  model_name            Name for the trained model.
  site                  Name of the cancer site for conditioning.

optional arguments:
  -h, --help            show this help message and exit

params_filepath could point to examples/IC50/example_params.json, examples for other files can be downloaded from here.

References

If you use paccmann_generator in your projects, please cite the following:

@article{born2021datadriven,
  author = {Born, Jannis and Manica, Matteo and Cadow, Joris and Markert, Greta and Mill, Nil Adell and Filipavicius, Modestas and Janakarajan, Nikita and Cardinale, Antonio and Laino, Teodoro and {Rodr{\'{i}}guez Mart{\'{i}}nez}, Mar{\'{i}}a},
  doi = {10.1088/2632-2153/abe808},
  issn = {2632-2153},
  journal = {Machine Learning: Science and Technology},
  number = {2},
  pages = {025024},
  title = {{Data-driven molecular design for discovery and synthesis of novel ligands: a case study on SARS-CoV-2}},
  url = {https://iopscience.iop.org/article/10.1088/2632-2153/abe808},
  volume = {2},
  year = {2021}
}

@article{born2021paccmannrl,
  title = {PaccMann\textsuperscript{RL}: De novo generation of hit-like anticancer molecules from transcriptomic data via reinforcement learning},
  journal = {iScience},
  volume = {24},
  number = {4},
  pages = {102269},
  year = {2021},
  issn = {2589-0042},
  doi = {https://doi.org/10.1016/j.isci.2021.102269},
  url = {https://www.cell.com/iscience/fulltext/S2589-0042(21)00237-6},
  author = {Born, Jannis and Manica, Matteo and Oskooei, Ali and Cadow, Joris and Markert, Greta and {Rodr{\'{i}}guez Mart{\'{i}}nez}, Mar{\'{i}}a}
}