Distributed representations of graphs for drug pair scoring

TL;DR

This paper investigates the use of distributed graph representations in drug pair scoring, demonstrating their effectiveness across various models and tasks, and releases these embeddings publicly.

Contribution

It introduces a methodology for learning and integrating distributed graph representations into drug pair scoring models, improving their performance.

Findings

Embeddings improve performance across multiple models and tasks.

Distributed representations are effective despite dataset growth and updates.

Public release of drug embeddings for several datasets.

Abstract

In this paper we study the practicality and usefulness of incorporating distributed representations of graphs into models within the context of drug pair scoring. We argue that the real world growth and update cycles of drug pair scoring datasets subvert the limitations of transductive learning associated with distributed representations. Furthermore, we argue that the vocabulary of discrete substructure patterns induced over drug sets is not dramatically large due to the limited set of atom types and constraints on bonding patterns enforced by chemistry. Under this pretext, we explore the effectiveness of distributed representations of the molecular graphs of drugs in drug pair scoring tasks such as drug synergy, polypharmacy, and drug-drug interaction prediction. To achieve this, we present a methodology for learning and incorporating distributed representations of graphs within a unified framework for drug pair scoring. Subsequently, we augment a number of recent and state-of-the-art models to utilise our embeddings. We empirically show that the incorporation of these embeddings improves downstream performance of almost every model across different drug pair scoring tasks, even those the original model was not designed for. We publicly release all of our drug embeddings for the DrugCombDB, DrugComb, DrugbankDDI, and TwoSides datasets.