Fast and robust graph construction from KEGG metabolic and genomic data

The Kyoto Encyclopedia of Genes and Genomes (KEGG) enables the modelling of biological systems as integrated graphs, where directed metabolic networks capture reaction flows and undirected genomic graphs encode chromosomal proximity of enzyme coding genes. However, constructing these graphs at the scale of complete bacterial organisms remains challenging due to two limitations: web based API access is constrained by rate limits triggering temporary IP blocks, and existing tools are typically restricted to the scale of pathway analysis. Here we present a framework for the fast and robust construction of whole organism metabolic and genomic graphs from KEGG data. Our approach combines a concurrency model with proxy rotation to overcome web access bottlenecks, specialised parsers for tab-delimited and KGML formats, and an embedded analytical database for efficient storage. For graph construction, we introduce optimised methods that handle real world data complexities, including multiple genomic intervals per gene and uncertain boundary annotations. This enables, to the best of our knowledge, the first systematic construction of integrated graphs at the scale of complete bacterial genomes and entire metabolisms. By overcoming previous scalability barriers, our framework provides the foundation for applying advanced trail finding algorithms to reveal how metabolic function and genomic organisation co-evolve across the bacterial domain, moving beyond isolated pathways towards whole-organism comparative analysis.