Table 4 Support scientists would have liked for explanatory analysis

Content

Edges in networks weighted by biological traits

Overlays of protein-protein interactions and disease associations

Overlays of protein-protein interactions and relevant pathways

Distinctions between proteins and other molecules that might serve as mediators of interactions, e.g. enzymes

Test statistics and counts (e.g. # of interactions, # of articles, overrepresentation of a functional term) and perceptually encoding nodes or links by them

Interactivity

Updating of interactions (e.g. selection, color coding) across views – e.g. across overlaid networks

Facile filtering (users had to use mini-scripting to filter)

Facile color-coding (at the time it took 15+ steps to color code)

Integrating one's own data into the displayed dataset

Simplifying networks through zooming, filtering, color-coding, expanding some nodes but not others, mapping only select neighbors to pathways

Conducting computations on networks to find e.g. shared paths to identify indirect interactions or recurrent or aberrant patterns that might signal a biologically significant set of relationships

Workspaces

Spaces for comparing different networks side-by-side with dynamically linked interactions

Spaces for aggregating entities on the fly into manipulable qualitative attributes based on emerging knowledge (e.g. normal vs disease conditions)