MetaModule: Reconstruction and visualization of module-based metabolic networks

Metabolism comprises the chemical reactions that sustain life. These reactions form complex metabolic networks that are organized into metabolic pathways, each handling a particular metabolic function. At a finer granularity, pathways are further organized into KEGG modules, which consist of functional units that capture short sequences of reactions. The organization of metabolic networks into pathways and modules is provided by the Kyoto Encyclopedia of Genes and Genomes (KEGG) database, a reliable and widely used resource of metabolic data. Although KEGG modules offer interpretable small functional building blocks within pathways, few tools exist to visualize or analyze them systematically in a network. This thesis introduces MetaModule, a Python-based pipeline for reconstructing and exploring module-based metabolic networks from lists of KEGG Orthology (KO) identifiers. Specifically, MetaModule parses KEGG module definitions, evaluates module completeness based on the input KOs (e.g., an organism or metagenomic sample), and builds the corresponding Module Graph: a directed network where nodes are modules and edges reflect potential metabolic flow via shared compounds. The pipeline generates an interactive visualization of this graph, allowing users to explore the network with deterministic layouts that ensure reproducibility. A key feature is a comparative mode that enables the visual analysis of two KO sets, highlighting shared and distinct modules within the same stable layout. Since KEGG provides module pages but not structured, machine-readable diagrams formats, we also developed parsers that reconstruct both the module KO-definition structure and reaction lists. These enable automatic generation of two complementary diagram types: KO definition diagrams and reaction flow diagrams. Both are available as interactive and downloadable visualizations. MetaModule produces layouts that emphasize readability and scales to complete organism-level module networks. All outputs are bundled in a single structured JSON for reuse and integration into downstream analyses. MetaModule constitutes the computational core of a larger web application that is currently under development; this thesis focuses on the pipeline itself. Overall, this work provides a reproducible, module-based framework that complements existing pathway-focused approaches by reconstructing and exploring metabolic networks, and making comparative differences easy to inspect and interpret.