Benchmarking Accuracy and Generalizability of Four Graph Neural Networks Using Large In Vitro ADME Datasets from Different Chemical Spaces

TL;DR

This study benchmarks four graph neural network models against traditional machine learning methods on large ADME datasets, revealing GAT as a promising approach with comparable accuracy to experimental assays, highlighting the importance of model and data quality.

Contribution

It provides a comprehensive comparison of GNN variants on industrial ADME datasets, demonstrating GAT's superior performance and the impact of experimental error on model accuracy.

Findings

GAT outperforms other GNN variants and traditional models.

All GNNs significantly outperform fingerprint-based models.

Model accuracy is comparable to inter-laboratory experimental variability.

Abstract

In this work, we benchmark a variety of single- and multi-task graph neural network (GNN) models against lower-bar and higher-bar traditional machine learning approaches employing human engineered molecular features. We consider four GNN variants -- Graph Convolutional Network (GCN), Graph Attention Network (GAT), Message Passing Neural Network (MPNN), and Attentive Fingerprint (AttentiveFP). So far deep learning models have been primarily benchmarked using lower-bar traditional models solely based on fingerprints, while more realistic benchmarks employing fingerprints, whole-molecule descriptors and predictions from other related endpoints (e.g., LogD7.4) appear to be scarce for industrial ADME datasets. In addition to time-split test sets based on Genentech data, this study benefits from the availability of measurements from an external chemical space (Roche data). We identify GAT as a promising approach to implementing deep learning models. While all GNN models significantly outperform lower-bar benchmark traditional models solely based on fingerprints, only GATs seem to offer a small but consistent improvement over higher-bar benchmark traditional models. Finally, the accuracy of in vitro assays from different laboratories predicting the same experimental endpoints appears to be comparable with the accuracy of GAT single-task models, suggesting that most of the observed error from the models is a function of the experimental error propagation.