Saturday, December 17, 2005

Visualising the Tree of Life


This is an area which has received a lot of attention. For some examples see Rebecca Shapley's Tree of Life Gallery, and the information aesthetics post. I've used hyperbolic trees in my (now rather old) Glasgow Taxonomy Name Server.

To be honest, I've never felt that we've hit on a truly compelling way to visualise large trees. I'm particularly concerned that trees are sparse (there's a lot of empty space when drawing a tree), and it is often not obvious how to navigate around them. Contrast this with Google Maps, which has spawned masses of applications (see Google Maps Mania). I think maps are something everybody "gets", partly because you only move in two dimensions, and they are predictable (if I start in Glasgow and move west, I know that sooner or later I will hit the Atlantic Ocean). Furthemore, maps have a universal coordinate system, so that an item can be located by its latitude and longitude, and maps are visually dense (zoom in and more information appears - this is best seen on the wonderful Google Earth). In evolutionary and taxonomic trees, by contrast, there is no predictibility (if I move left what taxa will I encounter?), and no coordinate system (we can't easily add things to a tree, without of course, changing the tree).

Seems to be an area ripe for more work.

No comments: