Beyond Organic Molecules: MetalLipoDB, a Lipophilicity Dataset for Metal Complexes and Machine Learning Benchmarks
- 1N.S. Kurnakov Institute of General and Inorganic Chemistry, Russian Academy of Sciences, 119991 Leninskii pr. 31, Moscow, Russia
- 2Department of Chemistry, Lomonosov Moscow State University, Lenin’s hills 1/3, Moscow, Russia, 119991
Abstract
Lipophilicity governs the uptake, distribution and toxicity of drug candidates, yet the experimental logP datasets used to build predictive models contain almost exclusively organic molecules. Here we present MetalLipoDB, a manually curated dataset of 5158 experimental logP and logD values of complexes of 23 metals from 1192 articles, including 556 entries for multinuclear and heterometallic complexes, annotated with the measurement conditions and counterions. The dataset quantifies the inter-study reproducibility of these measurements (0.56 log units) and serves as a benchmark for machine learning, with a mean absolute error of 0.72–0.75 for complexes from unseen articles. For articles with at least four measurements, anchoring the predictions to three measured complexes of the series reduced the error on the remaining complexes from 0.73 to 0.51. MetalLipoDB, the code and an interactive web application (https://biometaldb.streamlit.app/) are openly available to support the design of metal-based drugs.
Related
- Accuracy and Efficiency Benchmarks of Pretrained Machine Learning Potentials for Molecular Simulations
- A machine learning-based platform for predicting the potential energy of small molecules
- ∆-Machine Learning for LC-DFT-level Excitation Energies of Bacteriochlorophyll Molecules in a LH2 Complex
- Machine-learning force-field scoring rivals free-energy perturbation for congeneric ligand ranking across public benchmarks
- Comparative Evaluation of Graph Neural Networks for Molecular Lipophilicity Prediction
BibTeX
@misc{krasnov2026beyond,
title = {{Beyond Organic Molecules: MetalLipoDB, a Lipophilicity Dataset for Metal Complexes and Machine Learning Benchmarks}},
author = {Lev Krasnov and Dmitriy Malikov and Marina A. Kiseleva and Artem Lvov and Sergei V. Tatarin and Stanislav I. Bezzubov},
howpublished = {ChemRxiv},
year = {2026},
month = oct,
doi = {10.26434/chemrxiv.15010021/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010021/v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Beyond Organic Molecules: MetalLipoDB, a Lipophilicity Dataset for Metal Complexes and Machine Learning Benchmarks
AU - Lev Krasnov
AU - Dmitriy Malikov
AU - Marina A. Kiseleva
AU - Artem Lvov
AU - Sergei V. Tatarin
AU - Stanislav I. Bezzubov
PY - 2026
DA - 2026/10/07
PB - ChemRxiv
DO - 10.26434/chemrxiv.15010021/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010021/v1
AB - Lipophilicity governs the uptake, distribution and toxicity of drug candidates, yet the experimental logP datasets used to build predictive models contain almost exclusively organic molecules. Here we present MetalLipoDB, a manually curated dataset of 5158 experimental logP and logD values of complexes of 23 metals from 1192 articles, including 556 entries for multinuclear and heterometallic complexes, annotated with the measurement conditions and counterions. The dataset quantifies the inter-study reproducibility of these measurements (0.56 log units) and serves as a benchmark for machine learning, with a mean absolute error of 0.72–0.75 for complexes from unseen articles. For articles with at least four measurements, anchoring the predictions to three measured complexes of the series reduced the error on the remaining complexes from 0.73 to 0.51. MetalLipoDB, the code and an interactive web application (https://biometaldb.streamlit.app/) are openly available to support the design of metal-based drugs.
ER -