paccmann_predictor

Drug interaction prediction with PaccMann.

paccmann_predictor is a package for drug interaction prediction, with examples of anticancer drug sensitivity prediction and drug target affinity prediction. Please see our papers:

• Toward explainable anticancer compound sensitivity prediction via multimodal attention-based convolutional encoders (Molecular Pharmaceutics, 2019). This is the original paper on IC50 prediction using drug properties and tissue-specific cell line information (gene expression profiles). While the original code was written in tensorflow and is available here, this is the pytorch implementation of the best PaccMann architecture (multiscale convolutional encoder).

PaccMann for affinity prediction:

• 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 propose a slightly modified version to predict drug-target binding affinities based on protein sequences and SMILES

Installation

The library itself has few dependencies (see setup.py) with loose requirements. First, set up the environment as follows:

conda env create -f examples/IC50/conda.yml
conda activate paccmann_predictor
pip install -e .

Evaluate pretrained drug sensitivty model on your own data

First, please consider using our public PaccMann webservice as described in the NAR paper.

To use our pretrained model, please download the model from: https://ibm.biz/paccmann-data (just download models/single_pytorch_model). For example, assuming that you:

1. Set up your conda environment as described above;
2. Downloaded the model linked above in a directory called single_pytorch_model and
3. Downloaded the data from https://ibm.box.com/v/paccmann-pytoda-data in folders data and splitted_data; then, the following command should work:

(paccmann_predictor) $ python examples/IC50/test_paccmann.py \
splitted_data/gdsc_cell_line_ic50_test_fraction_0.1_id_997_seed_42.csv \
data/gene_expression/gdsc-rnaseq_gene-expression.csv \
data/smiles/gdsc.smi \
data/2128_genes.pkl \
single_pytorch_model/smiles_language \
single_pytorch_model/weights/best_mse_paccmann_v2.pt \
results \
single_pytorch_model/model_params.json

NOTE: If you bring your own data, please make sure to provide the omic data for the 2128 genes specified in data/2128_genes.pkl. Your omic data (here it is data/gene_expression/gdsc-rnaseq_gene-expression.csv) can contain more columns and it does not need to follow the order of the pickled gene list. But please dont change this pickle file. Also note that this is PaccMannV2 which is slightly improved compared to the paper version (context attention on both modalities).

Finetuning on your own data

You can also finetune our pretrained model on your data instead of training a model from scratch. For that, please follow the instruction below for training on scratch and just set:

• model_path --> directory where the single_pytorch_model is stored
• training_name --> this should be single_pytorch_model
• params_filepath --> base_path/single_pytorch_model/model_params.json

Training a model from scratch

To run the example training script we provide environment files under examples/IC50/. In the examples directory is a training script train_paccmann.py that makes use of paccmann_predictor.

(paccmann_predictor) $ python examples/IC50/train_paccmann.py -h
usage: train_paccmann.py [-h]
                         train_sensitivity_filepath test_sensitivity_filepath
                         gep_filepath smi_filepath gene_filepath
                         smiles_language_filepath model_path params_filepath
                         training_name

positional arguments:
  train_sensitivity_filepath
                        Path to the drug sensitivity (IC50) data.
  test_sensitivity_filepath
                        Path to the drug sensitivity (IC50) data.
  gep_filepath          Path to the gene expression profile data.
  smi_filepath          Path to the SMILES data.
  gene_filepath         Path to a pickle object containing list of genes.
  smiles_language_filepath
                        Path to a pickle object a SMILES language object.
  model_path            Directory where the model will be stored.
  params_filepath       Path to the parameter file.
  training_name         Name for the training.

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_predictor in your projects, please cite the following:

@article{manica2019paccmann,
  title={Toward explainable anticancer compound sensitivity prediction via multimodal attention-based convolutional encoders},
  author={Manica, Matteo and Oskooei, Ali and Born, Jannis and Subramanian, Vigneshwari and S{\'a}ez-Rodr{\'\i}guez, Julio and Mart{\'\i}nez, Mar{\'\i}a Rodr{\'\i}guez},
  journal={Molecular pharmaceutics},
  volume={16},
  number={12},
  pages={4797--4806},
  year={2019},
  publisher={ACS Publications},
  doi = {10.1021/acs.molpharmaceut.9b00520},
  note = {PMID: 31618586}
}

@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}
}