Network Visualisation for huge dataset composed of many unconnected clusters I have a dataset which consists of 190,455 nodes and 1,241,638 edges. These can be broken down into a set of 2300 subgraphs which are not connected to each other. I am having trouble creating visualisations for this using tools such as Gephi (Yifan-Yu seems to work best) as the layout takes a very long time and doesn't seem to separate the subgraphs well.
Is there a tool which performs a layout for such a dataset well? It would seem best to plot each subgraph individually, and then perform some merging at the end to avoid overlap while the algorithms seem to be trying to plot and arrange everything at once.
Thanks!
 A: What you need to do is use a layout algorithm that recognizes the clusters and places them individually. There are two interactive network visualization tools I know of that will do that for you, though both will take quite a while to handle such a large network.
Cytoscape is a Java tool built mainly for biologists, but will automatically separate disconnected components into their own sections of the screen and lay them out individually, like in this image. Sometimes it isn't very space-filling.

NodeXL is a template for Excel 2007/2010 for network analysis that includes a new Group-in-a-Box layout that separates groups into their own region of the screen, whether the groups are components, clusters, or manually created. You would want to use this approach if you have any connections between your clusters at all. The interactive tool may be too slow for such a large network, but it does have a C# library you can use to run these algorithms yourself. Below are some examples and a reference for the associated paper. Disclaimer: I am an advisor and developer on the project.


*

*Rodrigues, E. M.; Milic-Frayling, N.; Smith, M.; Shneiderman, B. & Hansen, D. Group-in-a-Box layout for multi-faceted analysis of communities SocialCom '11: Proc. 2011 IEEE 3rd International Conference on Social Computing, 2011, 354-361. DOI:10.1109/PASSAT/SocialCom.2011.139



A: I've had good experiences with Graphviz, and while I've never drawn graphs that large myself, I know that it's supported. See this page and this page. 
