Recon3D
Summary¶
Brunk, Sahoo, Zielinski, et al. (2018) present Recon3D, a genome-scale reconstruction of human metabolism that, unlike prior Recon versions, integrates three-dimensional metabolite and protein structure data directly with the metabolic network — enabling structurally-resolved analysis of how disease mutations and drugs perturb metabolic function. At publication it was the most comprehensive human metabolic network model available, accounting for 3,288 open reading frames (17% of functionally annotated human genes), 13,543 metabolic reactions, 4,140 unique metabolites, and 12,890 protein structures.
Content and Method¶
- Network scale: 3,288 open reading frames, 13,543 metabolic reactions, 4,140 unique metabolites — an expansion from Recon1 (1,496 genes, 3,311 reactions) and Recon2 (1,675 genes, 7,785 reactions).
- 3D structural integration: protein structures from the Protein Data Bank (PDB) and metabolite structures from ChEBI are mapped onto the reaction network, with atom-level atom-atom transition mappings analyzed for 8,315 reactions, allowing mutations to be interpreted by their 3D spatial location rather than only their linear sequence position (mutations distant in sequence can be structurally proximal in the folded protein).
- Visualization: the first web-based visualization of 3D macromolecular structures in the context of their neighboring reactions, metabolites, and metabolic subsystems, built on the RCSB PDB website and an interactive "ReconMap" using the Google Maps API.
- Applications demonstrated: functional characterization of disease-associated mutations, and identification of metabolic response signatures caused by drug exposure (via a genetic-algorithm-based "IndiFinder" tool built on the COBRA Toolbox).
Role in the Metabolic Modeling Ecosystem¶
Recon3D is a foundational reconstruction referenced throughout downstream genome-scale metabolic modeling work: the Multi-scale Erythrocyte Metabolism Platform built its GEM-PRO (genome-scale model with protein structures) methodology in the same spirit for a single-cell-type model, and the organ-specific flux-mapping framework described in Metabolic Flux Modulation of Genetic Risk built its Harvey/Harvetta multi-organ models directly on top of Recon3D before lifting them over to the newer HUMAN1 reconstruction (finding that HUMAN1-based and Recon3D-based flux maps only partially agree, attributed to Recon3D's less complete gene-reaction annotation).
Availability¶
Web resource: vmh.life and bigg.ucsd.edu. Structure visualization: rcsb.org. Data and code: github.com/SBRG/Recon3D.
See Also¶
- Metabolic Flux Modulation of Genetic Risk — downstream organ-specific flux-mapping framework built on Recon3D-derived multi-organ models.
- Multi-scale Erythrocyte Metabolism Platform — an earlier, cell-type-specific GEM-PRO (structure-integrated metabolic model) for the human erythrocyte, using the same structural-integration philosophy.