THEMAP: Task Hardness Estimation for Molecular Activity Prediction¶
THEMAP is a Python library for computing distances between chemical datasets and estimating task hardness for bioactivity prediction. It helps researchers identify similar tasks for transfer learning and quantify prediction difficulty.
-
Quick Start
Get up and running with THEMAP in minutes
-
Tutorials
Step-by-step guides for common workflows
-
API Reference
Detailed documentation for all modules
Key Features¶
- Multi-modal distance computation: Molecular (OTDD, Euclidean, Cosine), protein (ESM2 embeddings), and metadata distances
- 27 molecular featurizers: Fingerprints, descriptors, and neural embeddings
- Production-ready: Caching, parallel processing, GPU acceleration
- CLI and Python API: Use from the command line or programmatically
- Unified Task system: Molecules, proteins, and metadata in one framework
Installation¶
# From PyPI (recommended)
pip install themap
# With all optional dependencies (ML, protein, OTDD)
pip install "themap[all]"
To install from source for development:
git clone https://github.com/HFooladi/THEMAP.git
cd THEMAP
source install.sh # uv-based; creates .venv and installs dev extras
See Getting Started for the optional dependency groups.
Quick Examples¶
Command Line¶
# Compute distances between datasets
themap quick datasets/ -f ecfp -m euclidean -o output/
# Full pipeline with config file
themap run config.yaml
Python One-Liner¶
from themap import quick_distance
results = quick_distance(
data_dir="datasets",
output_dir="output",
molecule_featurizer="ecfp",
molecule_method="euclidean",
)
# Results saved to output/molecule_distances.csv
Programmatic API¶
from themap import DatasetDistance, DatasetLoader
# Load datasets
loader = DatasetLoader("datasets/")
train_datasets = loader.load_fold("train")
test_datasets = loader.load_fold("test")
# Compute distance matrix
dd = DatasetDistance(
train_datasets=train_datasets,
test_datasets=test_datasets,
featurizer="ecfp",
method="euclidean",
)
distance_matrix = dd.compute()
Performance¶
| Method | Speed | Memory | Best For |
|---|---|---|---|
| OTDD | Slower | High | Small-medium datasets, highest accuracy |
| Euclidean | Fast | Low | Large datasets |
| Cosine | Fast | Low | High-dimensional features |
Citation¶
If you use THEMAP in your research, please cite:
@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},
volume={64},
number={10},
pages={4031-4046},
year={2024},
publisher={ACS Publications}
}
License¶
THEMAP is released under the MIT License. See LICENSE for details.