Modular core network constructed from Escherichia coli transcriptome datasets using a hypergraph-based pan-network approach

We developed a hypergraph-based pan-network framework to integrate transcriptomic co-expression networks across diverse conditions. By applying frequent itemset mining to dataset-specific gene clusters, we constructed a hypergraph capturing frequently co-expressed gene sets. Extracting high-frequency hyperedges yielded a robust core modular network. Applied to 106 Escherichia coli transcriptomes, our method accurately recapitulates 70% of known operons within its core modules, outperforming conventional graph-based approaches. Furthermore, through inter-modular network visualization and modularity profiling, this approach successfully reveals the dynamic reorganization of bacterial co-expression architectures in response to environmental changes, offering a powerful tool for large-scale omics integration.