Integration of graph neural networks and genome-scale metabolic models for predicting ...

Summary

The article discusses a graph neural network model called FlowGAT, created to predict gene essentiality using metabolic reaction graphs derived from Flux Balance Analysis (FBA) solutions. The model utilizes an attention mechanism within a GNN framework for binary classification of gene essentiality based on metabolic network connectivity and flow-based features. The tags highlight the marriage of graph theory, signal-flow graphs, bioinformatics, and systems biology for the purpose of assessing gene function.