Pretraining Graph Transformers with Atom-in-a-Molecule Quantum Properties for Improved ADMET Modeling

TL;DR

This study demonstrates that pretraining Graph Transformers with atom-level quantum properties enhances ADMET property prediction accuracy, with analysis revealing improved atomic environment representations and model interpretability.

Contribution

It introduces a novel pretraining approach using atomic quantum properties for Graph Transformers, outperforming other strategies in ADMET modeling.

Findings

Pretraining with atomic quantum properties yields better ADMET prediction results.

Models retain pretraining information after fine-tuning, affecting latent representations.

Pretraining influences the capture of atomic environment features in the model.

Abstract

We evaluate the impact of pretraining Graph Transformer architectures on atom-level quantum-mechanical features for the modeling of absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties of drug-like compounds. We compare this pretraining strategy with two others: one based on molecular quantum properties (specifically the HOMO-LUMO gap) and one using a self-supervised atom masking technique. After fine-tuning on Therapeutic Data Commons ADMET datasets, we evaluate the performance improvement in the different models observing that models pretrained with atomic quantum mechanical properties produce in general better results. We then analyse the latent representations and observe that the supervised strategies preserve the pretraining information after finetuning and that different pretrainings produce different trends in latent expressivity across layers. Furthermore, we find that models pretrained on atomic quantum mechanical properties capture more low-frequency laplacian eigenmodes of the input graph via the attention weights and produce better representations of atomic environments within the molecule. Application of the analysis to a much larger non-public dataset for microsomal clearance illustrates generalizability of the studied indicators. In this case the performances of the models are in accordance with the representation analysis and highlight, especially for the case of masking pretraining and atom-level quantum property pretraining, how model types with similar performance on public benchmarks can have different performances on large scale pharmaceutical data.