Aug 2026· Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2· pp. 5512-5523· 0 citations· 43 references
TL;DR
MuSeL is proposed, a multi-scale adaptive graph representation learning framework that jointly mitigates the above issues from two complementary perspectives: global topology modeling and local structure optimization.
Abstract
Microbe-drug association (MDA) prediction is of great importance for understanding drug action mechanisms and exploring microbiome-based therapeutic strategies. However, when confronted with extremely sparse biological networks with pronounced structural heterogeneity, existing methods often struggle to simultaneously model global topological dependencies and local structural disparities. To address these challenges, we propose MuSeL, a multi-scale adaptive graph representation learning framework that jointly mitigates the above issues from two complementary perspectives: global topology modeling and local structure optimization. Specifically, we develop a spectral kernel attention mechanism that leverages the normalized Laplacian and chebyshev polynomial expansion to efficiently capture multi-scale global topological relations in the spectral domain. Meanwhile, we construct a structure-aware self-adaptive sampling module that dynamically adjusts neighborhood sampling based on node clustering coefficients and degree centrality, thereby improving the reliability of local structural and feature representations. Finally, the fused multi-scale node embeddings are fed into a prediction module to estimate MDA scores. Extensive experiments show that outperforms existing mainstream models, while ablation studies and case analyses further validate its effectiveness and robustness. Overall, MuSeL provides a practical computational framework for prioritizing candidate MDAs, supporting downstream biological validation and drug repurposing research.
Chemoresistance is a major contributor to cancer treatment failure, and microRNAs (miRNAs) play a critical role in mediating this resistance by regulating gene expression. Therefore, identifying miRNA-drug associations is of great significance for advancing cancer therapy. However, existing computational models face significant challenges, including heterogeneous feature integration and data sparsity. To overcome these limitations, we propose a novel Multi-view Graph Learning Framework with Spectral Encoding and Sparse Cross-Attention (MVGSCA) for predicting miRNA-drug associations. The model constructs node features based on miRNA sequence similarity and drug SMILES similarity. Then it builds two distinct graphs: a gene-mediated functional graph from miRNA-drug target interactions and an association-guided structural graph from known miRNA-drug associations. These two graphs are linearly combined to produce a collaborative feature representation. To capture both local and global topological features, the model applies local power filtering and global heat kernel diffusion, followed by spectral encoding via Poisson-Charlier polynomial approximation to enhance the feature representation. Furthermore, a sparse cross-attention mechanism is introduced to dynamically weight and integrate heterogeneous features from multiple sources. On a benchmark dataset with 8,720 associations, MVGSCA achieves an AUC of 96.32% and an AUPR of 95.69% under five-fold cross-validation, significantly outperforming six state-of-the-art methods. Experimental results show that MVGSCA effectively integrates heterogeneous biological information and achieves superior prediction performance, offering valuable insights into cancer resistance mechanisms and supporting drug discovery efforts.
Ru Nie, Ying Fu, Zhengwei Li et al.· IEEE transactions on computa...· 0 citations
Identifying potential microbe–disease associations (MDAs) is vital for elucidating disease pathogenesis and advancing precision medicine. Existing methods primarily learn features from heterogeneous microbe–disease graphs, but often rely solely on global topology for feature propagation, ignoring neighborhood subgraph density, centrality, and edge-weight heterogeneity. The loss of such structural information further exacerbates distributional shifts of microbe–disease feature representations, making it difficult for static concatenation or average fusion to effectively bridge the semantic gaps between modalities or accurately capture the contextual dependencies between node pairs. To address these challenges, we propose STDCAMDA, a dual-channel learning framework for MDA prediction. For structural enhancement, we design a subgraph topology module that fuses global and local topological information via multidimensional edge weights and node-gating mechanisms, thereby encoding rare structural signals while suppressing noise. In feature learning, we adopt a dual-channel strategy: embedding dynamic weight correction into a graph convolutional network for adaptive adjacency calibration and building a cross-pooling attention network to mitigate modality distribution shifts and capture cross-modal dependencies. Finally, we introduce two strategies: a dynamically weighted fusion classifier that integrates dual-channel features and uses a multi-layer perceptron for prediction, and a subgraph-aware negative sampling strategy that selects hard negative samples. Experiments on the Disbiome and HMDAD datasets demonstrate that STDCAMDA outperforms seven existing MDA prediction models, with statistical significance tests confirming the reliability of these improvements and cold-start evaluations validating its robustness and generalization capability. Practically, STDCAMDA provides an effective computational framework for prioritizing candidate disease-related microbes and supporting downstream biomedical validation.
Jiahao Li, Xiangmin Ji, Xiaowen Gao et al.· Journal of King Saud Univers...· 0 citations
IHLO-DTI, a novel prediction model based on an improved hypergraph neural network and Laplacian matrix optimization, can effectively capture high-order many-to-many interactions between drugs and targets, improving prediction accuracy and robustness.
Guolongwei Dai, Tao Luo, Dandan Li et al.· ACS Synthetic Biology· 0 citations
SGTL-DDA is proposed, a novel graph transformer framework designed to incorporate structural information and domain-specific knowledge from heterogeneous biological information networks (HBINs) that successfully identifies both known therapeutics and novel repositioning candidates, supported by molecular docking results and literature evidence.
Bowei Zhao, Hui Zhao, Yu-an Huang et al.· IEEE transactions on computa...· 0 citations
MRGBMDAT, a multi-relational graph encoder network with bilinear fusion for miRNA-disease association type prediction significantly outperforms five state-of-the-art methods across multiple evaluation metrics, exhibiting strong discriminative power and generalization capability.
Y. Sun, Wenjing Su, Si-Qi Zhu et al.· Bioinform.· 0 citations
DC-MetaMG, a deep learning framework based on a causal disentanglement strategy that models association responses as the synergistic interplay between two complementary mechanisms: static binding and dynamic regulation, is proposed, demonstrating its suitability for training scenarios involving complex biological information.
Runzhou Tang, Xi Zhou, Yujie Qi et al.· IEEE journal of biomedical a...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
