Fig 1.
SBMLNetwork Layered Architecture.
The figure illustrates the modular, multi-layered design of SBMLNetwork, which is organized into discrete levels responsible for standard compliance, I/O operations, core processing, cross-language integration, and user interaction. In particular, the User API Layer is outlined with its three sub-layers that offer a seamless progression from high-level operations to granular control over visualization details and provide support for both novice and expert users.
Fig 2.
Performance Benchmark: SBMLNetwork vs. SBMLDiagrams Auto-layout.
Log–log plot of median wall-clock time for SBMLNetwork’s C++-based auto-layout engine (blue circles, solid fit) and SBMLDiagrams’ implementation of the pure-Python NetworkX spring_layout algorithm (red squares, dashed fit), applied to synthetic SBML models containing 20–2,000 species, with a fixed 4:1 species-to-reaction ratio. Each point represents the median of 10 runs (error bars are smaller than the markers). SBMLNetwork achieves a ∼240× speed-up for the smallest model and maintains a ∼18× advantage at the model with 2,000 species.
Fig 3.
Original vs. SBMLNetwork-generated repressilator pathway in SBGN format.
The left panel shows the original repressilator pathway from [24]; the right panel shows the same pathway rendered by SBMLNetwork, which embeds SBGN-compliant visualization data directly into the SBML model. This comparison highlights SBMLNetwork’s capability to generate high-quality, SBGN-compliant visualizations for models featuring distinct geometric shapes for reaction centers, explicit depiction of empty species, and flexible arrowhead representations for reaction curves.
Fig 4.
Original vs. SBMLNetwork-generated iNOS pathway in SBGN format.
The left panel shows the original iNOS pathway from [24]; the right panel shows the same pathway rendered by SBMLNetwork, which embeds SBGN-compliant visualization data directly into the SBML model. This comparison highlights SBMLNetwork’s capability to generate high-quality, SBGN-compliant visualizations for models with multiple compartments, multi-shaped and multi-labeled species, and multi-segmented reaction curves.
Fig 5.
Applying Escher-Style Template on the Core E. coli Metabolic Model.
This figure demonstrates the capability of SBMLNetwork to apply a style template to a complex biological model. The core E. coli metabolic model [26] was chosen, and its SBML Layout data was generated using EscherConvertor [1]. Subsequently, SBMLNetwork was employed to create SBML Render data and apply the well-known Escher style on the model. The resulting visualization showcases the advanced styling capabilities of SBMLNetwork, ensuring a uniform and fine-grained representation of complex biological networks in specific styles.
Fig 6.
Reaction Alignment of the Tricarboxylic Acid (TCA) Cycle Network.
Generated using the Python script in S1 Listing, SBMLNetwork’s arrangement feature positioned the pyruvate dehydrogenase reaction vertically at the top and arranged the core TCA-cycle reactions in a circular layout to emphasize pathway flow and its cyclic nature. Metabolites are shown in yellow, cofactors in blue, and reactions in green to distinguish the different components of the pathway.
Fig 7.
Color-encoded reaction fluxes in the S. cerevisiae glycolysis pathway.
SBMLNetwork maps reaction-flux values from the glycolysis pathway in a model of Saccharomyces cerevisiae [28] to a continuous color gradient—vivid hues for high fluxes and muted tones for low, which offers a visual snapshot of pathway activity at a selected simulation time point.