IEEE TSIPN Special Issue on Learning on Graphs for Biology and Medicine

Manuscript Due: 1 September 2024
Publication Date: March 2025

Networks or graphs are pervasive in biology and medicine, capturing complex interactions  such as molecular interaction maps, signaling transduction pathways, functional connectivity brain  networks, patient- disease-drug relationships among others. Moreover, in most biological networks, the  nodes are associated with attributes, which can be seen as signals on the nodes of the graph. Extracting  information from such complex systems, by exploiting the interplay between nodes and signal values, is  key towards understanding biological mechanisms and eventually designing data-driven approaches for  personalized medicine. However, dealing with such complex and irregular structures poses significant  challenges on the analytics side, and it requires the development of new signal processing and machine  learning tools that reveal complex relationships at different scales, and take into consideration inductive  biases coming from the application domain.

This special issue will highlight novel research that aims at designing methods that incorporate signal  processing, data geometry and topology to enable new scientific discoveries and better inference tools  for biomedical applications. Our goal is to bring together scientists from the signal processing community working on graph-based methods and encourage the application of these models to difficult biological or  clinical tasks within a variety of biomedical data contexts. We particularly welcome submissions on  methodological advancements in graph signal processing and learning that are inspired and guided by  biomedical challenges.

Some topics of interest include (but are not limited to) novel algorithms for biomedical  applications on methodological aspects such as:

Topics of Interest

  • Extension of classical signal processing notions (e.g., graph filtering, graph transforms) to graph-structured data for biomedical data inference and understanding
  • Higher-order graph representations for biomedicine
  • Graph representation learning including graph neural networks for biomedical applications
  • Joint/Multiview graph learning for subtype identification
  • Inference on multilayer networks for robust and invariant representations
  • Learning over heterogeneous graphs (e.g., knowledge graphs) for extracting knowledge from multimodal biomedical data
  • Generative graph models for new biological discoveries
  • Combining mechanistic knowledge from biology with graph representation learning for better inference
  • Foundation models on graphs for biomedicine

We encourage submissions across various data modalities (e.g., single cell gene expression, protein interaction networks, molecular networks, brain networks, biological knowledge graphs, medical imaging) and application domains such as disease understanding (e.g., protein structure prediction), precision medicine (e.g., analysis of gene expression, single-cell transcriptomics and multi-omics data), novel therapeutical development and antibiotic discovery (e.g., drug-drug and/or drug-target interaction prediction), neuroscience (e.g., combining functional and structural connectivity).

Important Dates

  • Manuscripts due: September 1, 2024
  • Review results and decision notification: November 1, 2024
  • Revised manuscripts due: December 1, 2024
  • Final acceptance notification: January 1, 2025
  • Camera-ready paper due: February 1, 2025
  • Publication date: March 2025

Guest Editors