Path-Augmented Graph Transformer Network

TL;DR

The paper introduces Path-Augmented Graph Transformer Networks (PAGTN), which incorporate path features to capture long-range dependencies in molecular graphs, outperforming traditional GCNs in various molecular property prediction tasks.

Contribution

It presents a novel graph transformer model that explicitly models long-range dependencies using path features, improving molecular property prediction accuracy.

Findings

PAGTN outperforms GCNs on multiple molecular datasets.

The model effectively captures higher-order graph properties.

Results demonstrate improved prediction accuracy across chemistry domains.

Abstract

Much of the recent work on learning molecular representations has been based on Graph Convolution Networks (GCN). These models rely on local aggregation operations and can therefore miss higher-order graph properties. To remedy this, we propose Path-Augmented Graph Transformer Networks (PAGTN) that are explicitly built on longer-range dependencies in graph-structured data. Specifically, we use path features in molecular graphs to create global attention layers. We compare our PAGTN model against the GCN model and show that our model consistently outperforms GCNs on molecular property prediction datasets including quantum chemistry (QM7, QM8, QM9), physical chemistry (ESOL, Lipophilictiy) and biochemistry (BACE, BBBP).