Benchmark on Drug Target Interaction Modeling from a Drug Structure Perspective

TMLR Paper5779 Authors

Abstract: The prediction modeling of drug-target interactions is crucial to drug discovery and design, which has seen rapid advancements owing to deep learning technologies. Recently developed methods, such as those based on graph neural networks (GNNs) and Transformers, demonstrate exceptional performance across various datasets by effectively extracting structural information. However, the benchmarking of these novel methods often varies significantly in terms of hyperparameter settings and datasets, which limits algorithmic progress. In view of these, we conducted a comprehensive survey and benchmark for drug-target interaction modeling from a structural perspective via integrating tens of explicit (i.e., GNN-based) and implicit (i.e., Transformer-based) structure learning algorithms. We conducted a macroscopical comparison between these two classes of encoding strategies as well as the different featurization techniques that inform molecules' chemical and physical properties. We then carry out the microscopical comparison between all the integrated models across the six datasets via comprehensively benchmarking their effectiveness and efficiency. To comprehensively assess fairness, we investigate model performance under two experimental scenarios: one with unified hyperparameter settings and the other with individually optimized configurations. Remarkably, the summarized insights from the benchmark studies lead to the design of model combos. We demonstrate that our combos can achieve new state-of-the-art performance on various datasets associated with cost-effective memory and computation.

Submission Length: Regular submission (no more than 12 pages of main content)

Previous TMLR Submission Url: https://openreview.net/forum?id=tYTZhB9AUU

Changes Since Last Submission:

1. We unified all dataset's early stopping patience to 50.
2. We use the hyperparameters from the respective publications for each method, and remove the unified hyperparameters setting.
3. We refined the data processing pipeline and ensured there is no information leakage caused by data source.
4. We rerun the experiments effected by modification and reported the new results.
5. We update Figure 4 with our latest results, and the plot reflects the convergence situation under early stopping.
6. We emphasized the cross-validation setting and early stopping used in *Section 3. A Fair Benchmark Platform Setup*

Assigned Action Editor: ~Markus_Heinonen1

Submission Number: 5779