Finding the most potent compounds using active learning on molecular pairs
Open Access
- 27 August 2024
- journal article
- research article
- Published by Beilstein Institut in Beilstein Journal of Organic Chemistry
- Vol. 20 (1) , 2152-2162
Abstract
Active learning allows algorithms to steer iterative experimentation to accelerate and de-risk molecular optimizations, but actively trained models might still exhibit poor performance during early project stages where the training data is limited and model exploitation might lead to analog identification with limited scaffold diversity. Here, we present ActiveDelta, an adaptive approach that leverages paired molecular representations to predict improvements from the current best training compound to prioritize further data acquisition. We apply the ActiveDelta concept to both graph-based deep (Chemprop) and tree-based (XGBoost) models during exploitative active learning for 99 Ki benchmarking datasets. We show that both ActiveDelta implementations excel at identifying more potent inhibitors compared to the standard exploitative active learning implementations of Chemprop, XGBoost, and Random Forest. The ActiveDelta approach is also able to identify more chemically diverse inhibitors in terms of their Murcko scaffolds. Finally, deep models such as Chemprop trained on data selected through ActiveDelta approaches can more accurately identify inhibitors in test data created through simulated time-splits. Overall, this study highlights the large potential for molecular pairing approaches to further improve popular active learning strategies in low data regimes by enabling faster and more accurate identification of more diverse molecular hits against critical drug targets.Keywords
Funding Information
- NIH NIGMS (R35GM151255)
- Duke Science & Technology Initiative
- U.S. Department of War (National Defense Science & Engineering Graduate (NDSEG) Fellowship Program)
This publication has 44 references indexed in Scilit:
- Chemprop: A Machine Learning Package for Chemical Property PredictionJournal of Chemical Information and Modeling, 2023
- Improving ΔΔG Predictions with a Multitask Convolutional Siamese NetworkJournal of Chemical Information and Modeling, 2022
- Pairwise Difference Regression: A Machine Learning Meta-algorithm for Improved Prediction and Uncertainty Quantification in Chemical SearchJournal of Chemical Information and Modeling, 2021
- Transfer learning for solvation free energies: From quantum chemistry to experimentsChemical Engineering Journal, 2021
- Adaptive Optimization of Chemical Reactions with Minimal Experimental InformationCell Reports Physical Science, 2020
- Artificial intelligence in chemistry and drug designJournal of Computer-Aided Molecular Design, 2020
- Practical considerations for active machine learning in drug discoveryDrug Discovery Today: Technologies, 2019
- Analyzing Learned Molecular Representations for Property PredictionJournal of Chemical Information and Modeling, 2019
- Selection of Informative Examples in Chemogenomic Datasets2018
- Active-learning strategies in computer-assisted drug discoveryDrug Discovery Today, 2015