Cellular Interplay in COVID-19: Insights from Graph Neural Networks with Multidimensional Edge Features
Primary Area: applications to physical sciences (physics, chemistry, biology, etc.)
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Keywords: Graph neural network, Classification, Biological application, Single cell RNA-seq, Cell-Cell Interaction
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.
Abstract: The COVID-19 has emerged as a global pandemic, posing a significant public health threat with its widespread infection and the potential for severe respiratory complications. Among the various methodologies employed, single-cell omics-based studies have been at the forefront, concentrating on “intra”-cellular properties exhibited by gene expression. However, given its infectious nature, complex biological processes, such as immune responses performed between immune cells, infected cells, etc., necessitate a deeper analysis on “inter”-cellular properties exhibited by cell-cell interaction scores calculated using ligand and receptor expression information. The differences in these interactions in addition to gene expression between severe and non-severe cases could be pivotal in understanding the disease’s onset and progression, including mechanism leading to disease severity. Since the structure representing the overall nature of immune response can be implemented by directed graph with cell types as nodes and their interactions as edges, we employed a Graph Neural Network (GNN) model architecture accommodating multi-dimensional edge features, one of the first applications in biological context. In this study, our model incorporates edge features of cell-cell interaction scores, and node features of transcriptional factors and their target genes, which are “intra”-cellular features affected in the downstream by “inter”-cellular features. By leveraging the power of GNNs and the innovative use of multiple edge features, our model offers a groundbreaking perspective on the biological complexity of COVID-19, holding promise for the development of more effective treatments and preventive measures.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 3279
Loading