Drug–drug interaction prediction with learnable size-adaptive molecular substructures

• Published in: Briefings in bioinformatics Vol. 23; no. 1
• Main Authors: Nyamabo, Arnold K, Yu, Hui, Liu, Zun, Shi, Jian-Yu
• Format: Journal Article
• Language: English
• Published: England Oxford University Press 17.01.2022; Oxford Publishing Limited (England)
• Subjects: Chemical composition; Drug interaction; Drug Interactions; Drug-Related Side Effects and Adverse Reactions; Drugs; Graph representations; Graphical representations; Humans; Message passing; Neural networks; Neural Networks, Computer; Predictions; Software; Source code; Substructure extraction; Gated Message Passing Neural Network; Drug–Drug Interaction; Multi-type interactions; Substructure interaction
• ISSN: 1467-5463; 1477-4054; 1477-4054
• DOI: 10.1093/bib/bbab441

Abstract Drug–drug interactions (DDIs) are interactions with adverse effects on the body, manifested when two or more incompatible drugs are taken together. They can be caused by the chemical compositions of the drugs involved. We introduce gated message passing neural network (GMPNN), a message passing neural network which learns chemical substructures with different sizes and shapes from the molecular graph representations of drugs for DDI prediction between a pair of drugs. In GMPNN, edges are considered as gates which control the flow of message passing, and therefore delimiting the substructures in a learnable way. The final DDI prediction between a drug pair is based on the interactions between pairs of their (learned) substructures, each pair weighted by a relevance score to the final DDI prediction output. Our proposed method GMPNN-CS (i.e. GMPNN + prediction module) is evaluated on two real-world datasets, with competitive results on one, and improved performance on the other compared with previous methods. Source code is freely available at https://github.com/kanz76/GMPNN-CS.