GNNs for Network Classification in Single Cell RNA Sequencing Data

Authors

Reid C. Sewell, Mississippi State UniversityFollow

ORCID

Reid C Sewell: https://orcid.org/0009-0005-9364-6752

MSU Affiliation

Data Science Academic Institute

Advisor

Dr. Andy Perkins

Abstract

A common technique when investigating a disease is to profile gene expression, as this gives unique insights into the functions of a cell. Gene expression data gathered from single cell RNA sequencing can be encoded into a gene co-expression network, which is a graph of potential relationships between different genes. One method for interpreting data encoded as a graph is to use a graph neural network, or GNN. This project designs and implements a GNN architecture to accomplish classification tasks on graph data. Then, given a dataset of gene co-expression networks made from multiple single cell RNA sequencing studies, the model architecture is used to train a GNN for classifying the health state a given gene co-expression network represents. The trained GNN model is used to judge the effectiveness of the architecture across disciplines

DOI

https://doi.org/10.54718/OJNC4917

Publication Date

5-12-2025

Keywords

GNN, Graph Neural Network, Network Classification, Single Cell RNA Sequencing

Disciplines

Computer Sciences | Data Science

Recommended Citation

Sewell, Reid C., "GNNs for Network Classification in Single Cell RNA Sequencing Data" (2025). Capstone Projects. 3.
https://scholarsjunction.msstate.edu/data-capstone/3