Simple semantic zooming

[ugurdogrusoz]

On low zoom levels (should be parametric), we should not display finer details such as the following, with the hope that it will improve rendering performance:

• node labels
• node infoboxes (state variables and units of information)
• interaction cardinalities Optionally:
• members of complexes
• entire edges (at especially low zoom levels)

[ahmetcandiroglu]

@ugurdogrusoz @kinimesi Node labels, interaction cardinalities, and entire edges can be hidden. However, when members of complexes are hidden whole graph display changes temporarily as shown. Should we include this feature as it is?

[ugurdogrusoz]

@ahmetcandiroglu thanks, looks very nice. Some questions:

• How do you actually "hide" edges or complex members? Looks like when topology changes, geometry of some objects (compartment "plasma membrane") drastically change for some reason.
• Can we do something about info boxes?

[ugurdogrusoz]

Looking at the performance results, there is not consistent, significant improvement. So, let's give up on this for now at least. @ahmetcandiroglu Could you please create a Google sheet and put your latest performance results there and leave the URL here as a comment please. Make sure explain with a few sentences how each test was performed.

[ahmetcandiroglu]

Five different pathways are used for four different experiments as follows.

• Collapse all and expand all features of Newt are tested and passed time is measured.
• A cytoscape animation is written to move all elements to a specific point in canvas to estimate time takes to drag elements and tested.
• Zoom is gradually increased from 0.125 to 16 (min to max) and passed time is measured.

See performance results