I have been utilizing Cytoscape for a few days and find it very powerful. However, some its functionalities I am still trying to unlock. Here is a network I have graphed , but because of how dense it is - the edges are crossing over nodes they aren't connected to.
Is there any way this can be prevented? Here is an illustration of the problem I speak of.
•
written
by
thorerges •
Choose a different layout guaranteed to have no such overlaps, such as Attribute Circle, or tweak the force-directed layout parameters to make the gravitation parameter negative (to increase repulsion between nodes), and re-render.
•
written
by
Alex Reynolds
I guess that is the only method then. Thank you.
•
written
by
thorerges •
Another method is to use hive plots, as Mikhail suggests.
As an example, I used them here to show qualitative, at-a-glance differences between classes of links separated by cell-specificity: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3679407/figure/F2/
Drawing points on a circle is probably the "optimal" solution for keeping links in a network from overlapping, but even with a circular representation, I don't think there's an easy way to get around hairballs when you're dealing with large datasets, sorry.
Though I will note that a large set of networks can be compared at a glance, so that's a nice feature. Depends on what you're trying to communicate, I guess.
•
written
by
Alex Reynolds
You may also be interested in Hive plots, those are far more flexible than circular layout.
Looks cool, but these Open Source platforms are difficult to learn, not really looking forward to learning something else of this kind.
•
written
by
thorerges •
Another nice link is Mike Bostock's d3.js experiments with Krzywinski's hive plot technique: http://bost.ocks.org/mike/hive/
He includes a nice web demo that shows the effect of bundling links in categories, with adjustable parameters: http://mbostock.github.io/d3/talk/20111116/bundle.html
•
written
by
Alex Reynolds
Log in to answer this question.