Visualization improvements

[dkoslicki]

• Currently, the scaling, spacing, etc. are all manually hard-coded here. These should be able to adapt to the size of the tree automatically (so font sizes, space, etc. all change automatically if the tree is very small/large, dense/sparse, etc.)

• Currently, log scaling is automatically applied here and here.

  • A default needs to be picked (log is fine for now)
  • A command line option to profile_to_plot.py should allow for other scaling (eg: raw abundances, sqrt, log, etc.)

• When ground truth is present (see #2), the default is to have the disk diameters on the trees reflect the ground truth relative abundances (iirc) if present at a given tree node, and if not present, use the prediction profile relative abundances to set the tree diameter. The disk is then split into two pieces, with each piece reflecting the ratio between the true and predicted relative abundances. This is visually sub-optimal (for example, while a perfect prediction would imply that a disk is split evenly down the middle showing both true and prediction reported the same relative abundance, other cases are hard to visually understand). Will need to

  • try alternate visualization
    • maybe two disks side by size
    • maybe two little bars with length reflect relative abundance
    • maybe a single bar with two different colors, amount of color proportional to true/predicted relative abundance
    • etc.
  • pick a default, perhaps expose alternatives

• Make organism names optional (only at the leaves, at all nodes, at no nodes, etc)

  • perhaps explore auto-abbreviating long names (eg. "Gammaproteobacteria bacterium TMED95" -> "g. bacterium TMED95") following these conventions).

• Fix the current issue with names and disks overlapping

• In general, make it as pretty as possible : )