Distance between nodes about cytoscape.js-cose-bilkent HOT 5 CLOSED

The short answer is no. Spring embedders including cose-bilkent do apply repulsion forces on nodes to avoid overlaps and put some minimal distance between nodes. I can't see the whole picture in your example but the reason that some nodes (AND, OR, ProteinCoding, and HCP) are not nicely distanced from each other might be cause your graph is very dense and the algorithm cannot satisfy all the constraints.

from cytoscape.js-cose-bilkent.

@superbobry If your data is a DAG/tree -- which logical AND / OR statements usually are -- you can also try the Dagre layout. Dagre is designed just for DAGs.

When using CoSE, you may have to alter the options for the force values to get a nice result. What values work well can be very different for different graphs, so you usually have to experiment a bit.

from cytoscape.js-cose-bilkent.

@ugurdogrusoz, @maxkfranz thank you very much for the explanation. I will give Dagre a try.

from cytoscape.js-cose-bilkent.

Do you know of a layout where I specify measured distances between nodes (edge lengths)
and the layout tries to place nodes so as to minimize distortions to the specified edge lengths?
One application is where nodes are genes and distances are measured (DNA sequence or perhaps recombination probability).
I want a map that tries to reconcile the measured distances with minimal distortion.

from cytoscape.js-cose-bilkent.

Do you know of a layout where I specify measured distances between nodes (edge lengths)
and the layout tries to place nodes so as to minimize distortions to the specified edge lengths?
One application is where nodes are genes and distances are measured (DNA sequence or perhaps recombination probability).
I want a map that tries to reconcile the measured distances with minimal distortion.

@AllanDickerman

Layouts like FCOSE and Cola allow for this. Simply set the values affecting edge length appropriately (e.g. idealEdgeLength and edgeElasticity in FCOSE). The Cola demo sets edge length proportional to the inverse of weight (i.e. closer nodes tend to be more related). GeneMANIA does the same thing with COSE.

N.b. if you overconstrain the system (e.g. with a large, dense network), then the lengths won't correspond well to your metric.

from cytoscape.js-cose-bilkent.