Four use cases:
- ML reconstruction: (think Mesquite pie diagrams); you change from state 0 to state 1 along a branch. Rather than coloring the entire branch in the color of the descendant node, it's more realistic to have a gradient from one state color to the other (makes even more sense in continuous case).
- Stochastic character mapping: color changing multiple times along a branch.
- Time interval: branches and portions of branches before the KT boundary get a different color than those after (maybe you've fit a different diversification model before and after the event). These could all be done by allowing branches to be subdivided and different colors put on each small part (though a higher level function to do the gradient would be convenient).
- Adaptive text decoration. Imagine you're zoomed in on a clade and want to look at bootstrap values. For a figure in a paper, we're all used to having most of the branch labels placed properly, but some may need to be moved a little (i.e, if there is a place with many short edges, you'll need to spread out the numbers a bit). Having software do this automatically would be ideal (i.e., test to see if two labels overlap, if so, use a spring algorithm to choose the placement of just those labels).
Use cases from May 18 meeting
- Polyploidy, duplication, hybridization
- Georeferences, related data
- Exploratory visualization, links to other forms of visualization and analysis