Here's a link to an aritcle in New Scientist.
The article discusses how an ability to make fragments of microtubules move more or less precisely along teflon "tracks" may, with refinement be the first nanotechnological solution to theproblem of positioning nanomachine components.
When I first heard of it, though, another application sprang to mind. The article doesn's discuss switching trains from track to track. Indeed, a significant part of the article is spent discussing the problems of keeping these things ON track.
However, if a technique is worked out for creating switches that can switch these trains from track to track, that would be significant to build a working computer from this technology, as I've heard some people were able to do with macro-scale model trains. (Any references on this anybody?)
Now granted the trains move along at only 1 micrometer per second, but depending on the scale that memory elements etc. could be made at, this could still be a viable technology for not particularly fast, but highly, highly minaturized computers.