In the previous version of the Reactome website, we used a visual overview of all pathways, called the "Sky". It represented each reaction as a small arrow, about 1mm long. These arrows were organized spatially into patterns reminiscent of the way that pathways are drawn in the Boehringer-Mannheim map. The problem was, we ran out of space for all of the little arrows. You can see what the old "Sky" looked like here. A special version of the "Sky", called the "SkyPainter", allowed a user to paint the reaction arrows different colors according to supplied expression data.
In the current Reactome website, there are tools for performing Expression analysis and species comparison. The initial results of these analyses are presented as tables, with one row per Reactome pathway. We would like to provide a more visual overview of the entire set of Reactome pathways, using visual cues to tell the user about average expression level, number of orthologous proteins, or whatever.
We envisage using a network representation to show this information. Nodes would be pathways, arcs would be relationships between them. A node's size indicates the numbers of proteins contained by the pathway. Edges between pathways are used to show overlapping among pathways (may be protein overlapping or reaction overlapping, or perhaps even interactions); the edge weights are proportional to the size of overlapping. In this way, we may draw a complete graph. We can choose a certain level of cutoff for edges, so that the number of edges is visually manageable for the user.
Reactome has many thousands of pathways, but these are organized hierarchically. To keep the display manageable, without too many nodes and arcs, you should focus on top-level pathways as listed in the left tree in this page: http://www.reactome.org/entitylevelview/PathwayBrowser.html#DB=gk_current.
We would like to be able to give the user a choice of what goes into the arcs, but we would be happy if they initially simply indicate the number of reactions connecting a given pair of pathways. The coloring of the nodes would depend on the tool feeding into them. E.g. for expression analysis, we could use mean expression value. We may also want to vary the size of nodes, depending on, e.g. the number of proteins a given pathway contains.
The nodes should stay in more or less the same place all the time, to make it easy on the user. So, the positions of the nodes would be pre-defined by us, rather than being laid out arbitrarily.