1. Understanding the interface
2. Navigating
3. Interpreting Labels
4. Viewing Interaction Information
5. Viewing Interactor Information
6. Viewing starburst gene neighbors
7. General Notes
8. Browser Compatibility

The interface is comprised of 5 small preview panes on the left hand side, with the currently selected graph displayed to the right of the previews.

The top preview pane, the composite view, contains all vertices and edges found in the other graphs. Each vertex represents a HomoloGene.
The remaining panes provide previews for human, worm, fly and yeast graphs, respectively. Clicking on a preview pane will select the associated graph for display. Alternatively, clicking on the links at the top of the display will yield the same result.

Moving the mouse pointer over vertices in the currently displayed graph will pop up a menu at the pointer's location. This menu will contain links to information about the gene or HomoloGene at this vertex, as well as an option to "re-center" on this gene.

Re-centering will move that gene to the center of the display and show neighboring genes up to two degrees away. If the gene is not in the BIND database, however, re-centering will not be allowed.

At the midpoint of all edges (except in the composite graph) is a black icon. Clicking on this icon will open a new window. The newly opened window will display information about the interaction between the genes at the edge's end points.

In the popup menu at each vertex are links to resources containing information about the gene (or HomoloGene).

In the composite graph, the HomoloGene identifier is displayed immediately below the vertex. In all other graphs, the gene name is displayed. If the gene name is unavailable, the protein refseq accession is displayed.

Genes with many immediate singly-connected neighbors are labelled as "starbursts" by the system. This is determined by the starburst threshold value entered into the form. Neighboring genes that are connected only to the starburst gene are collapsed into the starburst vertex. Only those neighboring genes with multiple connections are still shown. This helps to simplify the presented graph.

Clicking "expand starburst" on the pop-up menu will allow the user to see all of the starburst gene's immediate neighbors. If the starburst gene is not in the BIND database, expanding the starburst will be disallowed. However, if the same starburst is shown on one of the other graphs, and the gene the starburst is centered around is in BIND, expansion will be allowed.

Complex graphs:

Before graphs are computed and rendered, a quick survey is performed to estimate how complex the displayed graphs will be. If the network is overly complex, graphs will take a long time to generate and, in the end, will be too cluttered to be of use. In these cases, the system will alert the user about which graphs will be too complicated to render. An option to continue without displaying these graphs will be given.

Projected Genes:

Genes for each HomoloGene are projected onto all species (human, worm, fly, yeast).

Yeast Networks:

Due to the highly connected nature of yeast genes, only genes that are immediately connected to the gene of interest are included in the graph computations. Graphs for the other species include genes up to two degrees away from the gene of interest.

Ulysses has been tested on several platforms with different web browsers.

The system is known to work properly with the following browsers: