Synthetic Data for More Accurate Deep Learning Models in Molecular Science: A Test Case of Protein-Ligand Binding Affinity Prediction

Deep learning models are data-hungry, and synthetic (artificial) data has been shown to be invaluable when data availability is low. While this has been demonstrated in certain technology areas, adopting such an approach is new in machine learning (ML) applications in chemistry, except for some pre-training tasks. In drug discovery, predicting binding energy between proteins and ligands is crucial. Many ML-based studies have been proposed to predict protein-ligand binding affinity using existing experimental data. However, it has been shown that these models suffer from inherent biases. Recent efforts have resulted in PLAS-20k, a synthetic dataset of multiple protein-ligand complex (PLC) conformations generated using molecular dynamics (MD) simulation as a viable option to be used along with existing experimental data to improve binding affinity prediction. For the binding affinity prediction task, we employ Pafnucy, a deep convolutional neural network, and we propose using multiple structures for each PLC from PLAS-20k to train it. We compare four different statistical and ML-based result-aggregation techniques. This work demonstrates the utility of dynamic datasets in enhancing binding affinity predictions, laying the foundations for future improvements in predicting similar protein properties by using synthetic datasets and more sophisticated models and methods. We propose that synthetic datasets from physics-based methods can significantly help develop more accurate data-driven methods.