THEMAP

Task Hardness Estimation for Molecular Activity Predcition

Installation

THEMAP can be installed using pip. First, clone this repository, create a new conda environment with the required packages, and finally, install the repository using pip.

conda env create -f environment.yml
conda activate themap

pip install --no-deps git+https://github.com/HFooladi/otdd.git  
pip install --no-deps -e .

Getting Started

Basic Usage

import os
from dpu_utils.utils.richpath import RichPath

from themap.data import MoleculeDataset
from themap.data.distance import MoleculeDatasetDistance

source_dataset_path = RichPath.create(os.path.join("datasets", "train", "CHEMBL1023359.jsonl.gz"))
target_dataset_path = RichPath.create(os.path.join("datasets", "test", "CHEMBL2219358.jsonl.gz"))
source_dataset = MoleculeDataset.load_from_file(source_dataset_path)
target_dataset = MoleculeDataset.load_from_file(target_dataset_path)

molecule_feaurizer = "gin_supervised_infomax"
source_features = source_dataset.get_dataset_embedding(molecule_feaurizer)
target_features = target_dataset.get_dataset_embedding(molecule_feaurizer)

Dist = MoleculeDatasetDistance(D1=source_dataset, D2=target_dataset, method="otdd")

Dist.get_distance()
>>> {'CHEMBL2219358': {'CHEMBL1023359': 7.074298858642578}}

Reproduce FS-Mol Experiments

For the FS-Mol dataset, moleuclar embedding for each assay (ChEMBL id) and also, chemical and protein distance have been calculated and deposited in the zenodo.

1. Download it from zenodo
2. Unzip the directory and place it into datasets such that you have the path datasets/fsmol_hardness

Then, you can go to the notebooks folder, and run the notebooks.

Development

Tests

You can run tests locally with:

pytest

Code style

We use ruff as a linter and formatter.

ruff check
ruff format

Documentation

You can build and run documentation server with:

mkdocs serve

Citation

If you find the models useful in your research, we ask that you cite the following paper:

@article{fooladi2024quantifying,
  title={Quantifying the hardness of bioactivity prediction tasks for transfer learning},
  author={Fooladi, Hosein and Hirte, Steffen and Kirchmair, Johannes},
  journal={Journal of Chemical Information and Modeling},
  year={2024},
  publisher={ACS Publications}
}