Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Customized Subgraph Selection and Encoding for Drug-drug Interaction Prediction

Haotong Du, Quanming Yao, Juzheng Zhang, Yang Liu, Zhen Wang

TL;DR

This study introduces extensive subgraph selection and encoding spaces that account for the diverse contexts of drug interactions in DDI prediction, and designs a relaxation mechanism that uses an approximation strategy to efficiently explore optimal subgraph configurations.

Abstract

Subgraph-based methods have proven to be effective and interpretable in predicting drug-drug interactions (DDIs), which are essential for medical practice and drug development. Subgraph selection and encoding are critical stages in these methods, yet customizing these components remains underexplored due to the high cost of manual adjustments. In this study, inspired by the success of neural architecture search (NAS), we propose a method to search for data-specific components within subgraph-based frameworks. Specifically, we introduce extensive subgraph selection and encoding spaces that account for the diverse contexts of drug interactions in DDI prediction. To address the challenge of large search spaces and high sampling costs, we design a relaxation mechanism that uses an approximation strategy to efficiently explore optimal subgraph configurations. This approach allows for robust exploration of the search space. Extensive experiments demonstrate the effectiveness and superiority of the proposed method, with the discovered subgraphs and encoding functions highlighting the model's adaptability.

Customized Subgraph Selection and Encoding for Drug-drug Interaction Prediction

TL;DR

This study introduces extensive subgraph selection and encoding spaces that account for the diverse contexts of drug interactions in DDI prediction, and designs a relaxation mechanism that uses an approximation strategy to efficiently explore optimal subgraph configurations.

Abstract

Subgraph-based methods have proven to be effective and interpretable in predicting drug-drug interactions (DDIs), which are essential for medical practice and drug development. Subgraph selection and encoding are critical stages in these methods, yet customizing these components remains underexplored due to the high cost of manual adjustments. In this study, inspired by the success of neural architecture search (NAS), we propose a method to search for data-specific components within subgraph-based frameworks. Specifically, we introduce extensive subgraph selection and encoding spaces that account for the diverse contexts of drug interactions in DDI prediction. To address the challenge of large search spaces and high sampling costs, we design a relaxation mechanism that uses an approximation strategy to efficiently explore optimal subgraph configurations. This approach allows for robust exploration of the search space. Extensive experiments demonstrate the effectiveness and superiority of the proposed method, with the discovered subgraphs and encoding functions highlighting the model's adaptability.

Paper Structure

This paper contains 47 sections, 10 equations, 6 figures, 8 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Works
  3. Subgraph-based Link Prediction
  4. GNN-based DDI Prediction
  5. Graph Neural Architecture Search
  6. Proposed Method
  7. Problem Formulation
  8. Search Space
  9. Subgraph Selection Space
  10. Subgraph Encoding Space
  11. Search Strategy
  12. Search Problem
  13. Relaxation of Subgraph Selection Space
  14. Subgraph Representation Approximation Strategy
  15. Robust Search Algorithm
  16. ...and 32 more sections

Figures (6)

  • Figure 1: The pipeline of subgraph-based methods includes subgraph selection and subgraph encoding. In this work, we focus specifically on searching for components within the red-dotted lines.
  • Figure 2: Comparison on convergence between the searched architectures by CSSE-DDI and human-designed methods.
  • Figure 3: Performance given different hyperparameter $\eta$.
  • Figure 4: Visualization of the searched subgraphs corresponding to the specific drug pairs.
  • Figure 5: The searched encoding functions on all benchmark datasets.
  • ...and 1 more figures

Theorems & Definitions (1)

  • Definition 1: Customized Subgraph-based Pipeline Search Problem