Document Type
Conference Paper
Publication Date
2026
DOI
10.1145/3807503.3819442
Publication Title
BCB '26: Proceedings of the 17th ACM International Conference on Bioinformatics: Computational Biology and Health Informatics
Pages
82 (6 pp.)
Conference Name
BCB '26: 17 ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, 30 June-3 July, 2026, Rende (CS), Italy
Abstract
We propose AttF-GNN, an attention-based graph fusion strategy for diseases classification and subtyping. In multiomics analysis, not all types of molecular data are equally relevant for disease subtyping and considering all modalities equally may obscure discriminative signals and limit the effectiveness of predictive models by overlooking modality-specific contributions. Therefore, we design an attention-based multimodal GraphSAGE framework that can automatically emphasize the modalities providing the most relevant information for classification. At first, we have constructed three graphs using mRNA, RNA-seq and DNA methylation modalities, and train each omics with individual GraphSAGE encoders. Next, a unified intersection graph is formed using an attention-based method that integrates edges and nodes from all modalities. The attention-based fusion module combines modality-specific embeddings by aggregating node-level information into a pooled modality summary to capture information across all modalities. Lastly, we pass the unified graph through another GraphSAGE layer, which performs downstream tasks such as disease classification and subtype detection. AttF-GNN is trained and evaluated on four datasets, namely TCGA-BRCA, TCGA-PRAD, TCGA-GBM and ROSMAP. Experimental results demonstrate that AttF-GNN depicts strong performance across four datasets, which reflects the consistency of the proposed framework. Moreover, this model provides a modality score indicating which modality has the greatest influence on disease subtyping and classification. Together with the attention-based modality fusion and graph learning, it provides an expressive solution for cancer classification and subtype detection in the precision oncology domain.
Rights
© 2026 Copyright held by the owner/authors.
This work is licensed under a Creative Commons Attribution 4.0 International (CC BY 4.0) License.
Original Publication Citation
Chakraborty, S., Adam, E., Stilwell, T., Riethman, H., Ranjan, D., & Rana, P. (2026). Attf-GNN: An attention-based multi-omics graph neural network with modality learning for disease subtyping. In BCB '26: Proceedings of the 17th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics (Article 82). Association for Computing Machinery. https://doi.org/10.1145/3807503.3819442
Repository Citation
Chakraborty, S., Adam, E., Stilwell, T., Riethman, H., Ranjan, D., & Rana, P. (2026). Attf-GNN: An attention-based multi-omics graph neural network with modality learning for disease subtyping. In BCB '26: Proceedings of the 17th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics (Article 82). Association for Computing Machinery. https://doi.org/10.1145/3807503.3819442
ORCID
0000-0003-0765-1060 (Chakraborty), 0000-0002-7548-4375 (Adam), 0009-0001-1179-4606 (Stilwell), 0000-0003-4626-4733 (Riethman), 0000-0002-8298-7093 (Ranjan), 0000-0001-9199-2479 (Rana)