network graph of selection

I have a naïve question about doing a quick and dirty selection from a linked dataset. I am talking about nodes and edges, in two notebooks. One is @mbostock’s patent suits notebook, where I want to be able to use and then pass it to a new network graph. So that I can create a second network graph where Apple or Microsoft is the source or target, without using aq or other query language…
This would be so cool to do, with this fork of @mbostock’s patent suits…

Another network graph has json data, in separate files for nodes and edges. How do I produce a second graph with just one star (say, Raj Kapoor) and his connections?
Bollywood Stars

Thanks so much!