98%
921
2 minutes
20
Deep graph learning models have recently been developed to learn from various graphs that are prevalent in describing and modeling complex systems, including those in bioinformatics. However, a versatile explanation method for uncovering the general graph patterns that guide deep graph models in making predictions remains elusive. In this paper, we propose DGX, a novel deep graph model explainer that generates explanatory graphs to explain trained, black-box deep graph models. Its effectiveness is demonstrated by producing multiple graphs that collectively encode the structural knowledge captured by the graph neural network on both synthetic and real graph data. Importantly, DGX can produce diverse explanations by generating a set of distinguishable graphs and can provide customized explanations based on prior knowledge or constraints specified by users. We apply DGX to explain a mutagenicity prediction model by exploring the underlying groups of mutagenic compounds, and we explain the model on brain functional networks by revealing the structural patterns that enable the model to differentiate autism spectrum disorder from healthy controls. These findings offer an effective, diverse, and customized approach to explaining the underlying mechanisms and enhancing the understanding of models learned from real graph data, particularly in fields such as biomedicine and bioinformatics.
Download full-text PDF |
Source |
---|---|
http://dx.doi.org/10.1109/TCBBIO.2025.3593917 | DOI Listing |
Mol Divers
September 2025
Laboratory of Molecular Design and Drug Discovery, School of Science, China Pharmaceutical University, Nanjing, 211198, China.
Drug absorption significantly influences pharmacokinetics. Accurately predicting human oral bioavailability (HOB) is essential for optimizing drug candidates and improving clinical success rates. The traditional method based on experiment is a common way to obtain HOB, but the experimental method is time-consuming and costly.
View Article and Find Full Text PDFChem Sci
August 2025
Engineering Research Center of Cell & Therapeutic Antibody (MOE), School of Pharmacy, Shanghai Jiao Tong University Shanghai 200240 China
Predicting Antibody-Antigen (Ab-Ag) docking and structure-based design represent significant long-term and therapeutically important challenges in computational biology. We present SAGERank, a general, configurable deep learning framework for antibody design using Graph Sample and Aggregate Networks. SAGERank successfully predicted the majority of epitopes in a cancer target dataset.
View Article and Find Full Text PDFComput Med Imaging Graph
August 2025
Beijing Huilongguan Hospital, Peking University Huilongguan Clinical Medical School, Beijing 100096, China. Electronic address:
Bipolar disorder (BD) is a debilitating mental illness characterized by significant mood swings, posing a substantial challenge for accurate diagnosis due to its clinical complexity. This paper presents CS2former, a novel approach leveraging a dual channel-spatial feature extraction module within a Transformer model to diagnose BD from resting-state functional MRI (Rs-fMRI) and T1-weighted MRI (T1w-MRI) data. CS2former employs a Channel-2D Spatial Feature Aggregation Module to decouple channel and spatial information from Rs-fMRI, while a Channel-3D Spatial Attention Module with Synchronized Attention Module (SAM) concurrently computes attention for T1w-MRI feature maps.
View Article and Find Full Text PDFMed Eng Phys
October 2025
College of Basic Medical Science, Shanxi University of Chinese Medicine, Jinzhong, 030619, Shanxi, China.
Pulse diagnosis holds a pivotal role in traditional Chinese medicine (TCM) diagnostics, with pulse characteristics serving as one of the critical bases for its assessment. Accurate classification of these pulse pattern is paramount for the objectification of TCM. This study proposes an enhanced SMOTE approach to achieve data augmentation, followed by multi-domain feature extraction.
View Article and Find Full Text PDFIEEE Trans Comput Biol Bioinform
September 2025
Deciphering the three-dimensional structure of proteins remains a grand challenge in biology and medicine, as it holds the key to understanding their biological functions and facilitating drug discovery. In this paper, we introduce DECIPHER (Deep Encoding of Cellular Interactions and Protein HiErarchical Representation), a novel deep graph learning framework for protein structure prediction. By representing proteins as graphs, where residues and atoms serve as nodes and their interactions form edges, we capture the intricate spatial relationships within these complex biomolecules.
View Article and Find Full Text PDF