Workbench visualization tools will present one or more pathways, laying them out in 2D or 3D, and providing navigation tools to pan and zoom, adjust detail levels, or overlay additional information.
The sections below show mockups of navigation and visualization techniques under design for the workbench.
The familiar planar layout arranges molecule nodes and interaction edges on a 2D plane. Pan-and-zoom navigation selects the region of the plane to show. Node and edge icons and colors may be controlled by arbitrary data attributes. Optional automatic layout algorithms provide an initial arrangement of nodes that can be later altered by hand.
Pathway zoom
Pathway pan
As pathways grow large, Workbench users may use multiple windows to view the same pathway at different zoom levels or pan locations. User-positioned markers flag regions of interest and provide annotation. Clicking on a flag opens a window and sets the pan and zoom values to center on the region. Menus of flagged regions provide quick navigation in large pathways.
>Multiple windows and flagged regions
A Workbench lens effect magnifies a region of interest within a window. Though real-world lenses show the same content larger, visualization’s virtual lens can show different content inside and outside the lens. The lens interior may show a pathway with added detail, annotation, molecule and interaction descriptions, or simulation parameters and results. The lens interior may also show a different pathway, enabling quick regional comparisons between orthologs, pathological and normal states, or different pathway hypotheses.
Virtual zoom and ortholog lenses
Modular pathway components encapsulate a pathway fragment, exposing only a public interface of connection points and simulation parameters – much like the interfaces defined for programming language functions and classes. Larger pathways that use these components remain independent of changes in module structure, as long as the module’s public interface remains unchanged. Workbench level-of-detail control presents modular pathways as boxes without interior detail. Zooming into a module reveals its internal structure.
Modular pathways and level-of-detail control
Groups of nodes may be annotated to mark pathway fragments of interest. Automatic algorithms can scan a network and annotate node clusters with related attributes. Annotations can be nested, color coded, and saved. Attributes may be associated with annotations and annotations turned on and off individually or based upon these attributes.
Annotated pathways
Using 3D graphics hardware found standard on most computers today, pathways can be laid out on a 3D plane. Flight navigation navigates the viewer across the pathway plane. Perspective effects enlarge nearby regions of interest. Distant features are smaller, but remain visible to provide context. For large diagrams, distant features can be dimmed by depth cueing to help highlight the closer region of interest.
Layout on a 3D plane
Other 3D effects are possible for zooming in on content of interest while distant regions remain visible, but smaller. A hyperbolic or fish-eye effect zooms central content closer, while smoothly reducing the size of distant content in an M.C. Escher-like effect. A cylindrical roller effect slides content side-to-side, zooming in on a central stripe of the pathway. In each case, distant content remains visible to provide context around the region of interest.
Fish-eye and roller effects
3D may be used to layer multiple pathways, each drawn at a different elevation. Intelligent node and edge layout algorithms use the layout of one pathway to steer the layout of another, producing pathways with similar appearance. When layered, nodes and edges common to multiple pathways align vertically, making similarities and differences easier to discern. Color-coding highlights common structure.
Layered pathways
The height of a node or edge above the plane provides a visual attribute that may be controlled by data. A terrain layout moves nodes vertically off the plane, setting heights based upon a molecule’s attributes, such as its type (protein, DNA, etc.), it’s role in the pathway, the researcher’s confidence level that the molecule is correctly modeled, or any other attributes available.
Terrain layout
Where nodes are assigned spatial regions (e.g., cell membrane, nucleus, etc.), nodes may be laid out region-by-region. Iconic representations mark regions and provide background shading behind nodes contained within. When region spatial characteristics are known (e.g., size, location, shape), 3D shapes may be defined, nested within each other and enclosing the relevant nodes. Transparency and light shading make a region visible without occluding its contents. In cases where very detailed spatial knowledge is available, nodes may be laid out within 3D volumetric models of the regions.
Spatial regions in 2D and 3D
Pathway simulations produce a time-line of concentration changes that may drive animations of visual attributes. Nodes highlight as concentrations increase, then decrease, tracking the flow of a signal through a pathway. Animation controls select the portion of the simulation to playback and the playback speed. Clicking on the time-line jumps the animation to the selected point in the simulation.
Animation highlighting concentration changes