Deep learning is accelerating drug discovery. However, current approaches are often affected by limitations in the available data, e.g., in terms of size or molecular diversity. Active learning is poised to be a solution for drug discovery in low-data regimes. In active learning, a model is updated iteratively by taking multiple smaller screening steps, instead of suggesting many molecules at once with a single model for traditional ‘one-shot’ screening. This iterative approach aims to improve models during the screening process and can adjust course along the way. Despite pioneering applications, it is currently unclear how active learning holds up to traditional approaches and what the best strategies are for prospective drug discovery applications. In this study, we combine six active learning strategies with two distinct deep learning methods and analyse their performance on three large-scale libraries. For each strategy, we investigate its capacity to discover bioactive hits and explore the chemical space. We found that active learning can achieve up to a six-fold improvement in hit retrieval compared to traditional methods in a low-data scenario. The strategy to acquire new molecules at each cycle resulted to be the primary driver of performance, and it determined the ‘molecular journey’ throughout screening cycles. Finally, we show that active learning can quickly compensate for a lack of molecular diversity in the starting set. These results advocate for the adoption of active deep learning to accelerate drug discovery in low-data regimes.
This repository contains all code to replicate our study. Feel free to try out anything you find here.
data_prep.py: Processes active and inactive compound data.clustering.py: Clusters compounds for sampling diversity.nn.py: Contains neural network models (MLP, GCN, etc.).screening.py: Core script for active learning cycles.utils.py: Utility functions for data handling and evaluation.main.py: Entry point for running experiments with customizable parameters.
Run python main.py with desired command-line arguments to start the active learning process. Ensure necessary data files (active and inactive .smi) are present.
Install dependencies from the provided env.yaml file.
conda env create -f env.yaml
This codebase uses Python 3.9 and depends on:
- PyTorch (2.0.1)
- PyTorch Geometric (2.3.1)
- RDKit (2023.3.2)
- Scikit-learn (1.3.0)
Traversing Chemical Space with Active Deep Learning. Derek van Tilborg and Francesca Grisoni. ChemRxiv, 2023. DOI: https://doi.org/10.26434/chemrxiv-2023-wgl32
All code is under MIT license.