Graph Representation Learning for Biomedical Applications

Graphs are a powerful abstraction of the structure and dynamics of complex systems, providing a solid theoretical framework for developing methods tackling relevant biomedical problems, for example, the modeling of 3D protein structures, enrichment of incomplete protein-protein interaction networks, protein automatic function prediction, construction of accurate disease classification models, drug discovery and repurposing, medical images analysis.

Biomedical data involves the use of complex multimodal and possibly heterogeneous graphs that have been leveraged to effectively model a wide range of different entities (e.g. atoms in molecular structures, protein-protein interactions, gene co-expression networks, regulatory networks, drug-target networks, cellular networks, healthcare knowledge, clinical, genetic and functional similarities among patients).

From this standpoint, Graph Representation Learning is attracting increasing interest from the biomedical network research community. This area has been further fueled by the emergence of Graph Neural Networks, specifically designed for the analysis of graphs using neural network tools and paving the way for the development of novel techniques.

This topical collection focuses on the research on graph representation methods, exploring both its theoretical and practical dimensions in addressing biomedical problems. We aim at gathering innovative research inspired by various biomedical problems featuring novel techniques for the development and application of Graph Representation Learning methods in biology and medicine.

We particularly welcome (but are not limited to) submissions on advancements in network representation and learning that target biomedical challenges, such as:

- Theoretical aspects of Graph Representation Learning (e.g. Foundations in knowledge representation, Structural Network analysis)

- Graph Representation Learning algorithms (e.g. Graph Embedding Techniques, Heterogeneous network representations, Multimodal/Multiview graphs, Generative graph models)

- Learning task on Graphs (e.g. node/link/graph property prediction, graph explainability and visualization techniques, graph generation)

- Deep Learning on Graphs (e.g. Graph Neural Networks)

- Applications of Graph Representation Learning (e.g. Drug discovery and re-purposing and response, Biomarker discovery, Functional genomics, Molecular interaction, Molecular structure, Biomedical image analysis, Analysis on single-cell and multi-omics data, Patient similarity networks, EHR and clinical data analysis, patients prognosis/diagnosis/recurrence prediction, general applications in Precision Medicine)

- Representation and algorithm on dynamical/temporal Graphs

- Network datasets and benchmarks

- Generative models for biomedical entities (e.g. generative LLM models)

- Evolutionary networks (e.g. phylogenetic, pangenomic)

This Collection supports and amplifies research related to SDG 9.

Keywords: graph representation learning, biomedicine, artificial intelligence, network medicine, heterogeneous network, multimodal data, node prediction, link prediction, graph visualization, precision medicine