High Performance of Gradient Boosting in Binding Affinity Prediction
This work addresses a key challenge in drug discovery by offering a more efficient and scalable method for binding affinity prediction, though it appears incremental as it builds on traditional machine learning approaches.
The paper tackled the problem of predicting protein-ligand binding affinity for drug discovery by combining protein-ligand interaction and graph-level features in gradient-boosted decision trees, resulting in improved performance over existing solutions.
Prediction of protein-ligand (PL) binding affinity remains the key to drug discovery. Popular approaches in recent years involve graph neural networks (GNNs), which are used to learn the topology and geometry of PL complexes. However, GNNs are computationally heavy and have poor scalability to graph sizes. On the other hand, traditional machine learning (ML) approaches, such as gradient-boosted decision trees (GBDTs), are lightweight yet extremely efficient for tabular data. We propose to use PL interaction features along with PL graph-level features in GBDT. We show that this combination outperforms the existing solutions.