help with using biofabric

[Nilanjan Bhattacharya]

Hi,

I am playing around with biofabric to analyze a social network.  I've put together some questions on what I want to analyze.  Can you tell me how to do these using biofabric:

Who are the ones with the most connections (outgoing, incoming)?
I think this can be seen directly.  Not sure of outgoing and incoming.

For a specific user, who are they connected to?
Do they have strong connections to one or more users?
Can I model connections with strength, i.e,. a is connected to b with strength 7, but a is connected to c with strength 8?
Are there users with mostly outgoing connections, mostly incoming connections?

I am looking for help on doing these.

Thanks,
- Nilanjan

[William Longabaugh]

Sorry for the delay in replying. I've been out of the office for the
last week.

Four things should be useful for what you are trying to do:

1) Using the shadow links display option is essential! That way, you can
see all the links for each user in one place. This allows you to assess
the most connected users at a glance.

2) Using an edge attribute file to layout edges. While the link groups
feature is great for creating separate edge wedges for different classes
of links, it's not going to work for dividing the inbound and outbound
links into separate wedges for each node. But you could, for example,
export the edge attribute file for the network, then use some scripting
to reorder the edges in the network for each node so that e.g. the
inbound edges are shown in one wedge, and the outbound edges in another.
Then use this modified edge attribute file to layout the network using
edge attributes. This approach should allow you to address your
inbound/outbound questions.

3) Take a look at the example I have in the gallery at:
http://www.biofabric.org/gallery/index.html#Minerals The idea is that
you can create multiple links between two nodes to convey edge strength.
In that example, the number of edges between two nodes is equal to the
log base 2 of the strength. Each edge is tagged with a different label
in the .sif import so BioFabric does not consider the multiple edges to
be duplicates that are eliminated. Not the most elegant solution, but
one that works OK for the moment.

4) For seeing who a specific user is connected to, you can tell that by
looking directly at the edge wedge when you have shadow links on. But if
you want to view it in isolation, create a submodel. Select the user
node of interest by clicking on it, hit the 1-degree button on the
toolbar to select first-degree neighbors and associated links, and click
the create submodel toolbar button (far right) to create the model and
view that one node in isolation.

Hope these tips help. My blog discusses many of these topics (but not
the edge attribute one, yet) if you need further info.

Bill

> --
> You received this message because you are subscribed to the Google
> Groups "BioFabric-users" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to biofabric-use...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.