Measurements panel improvements

[joverlee521]

Context
Improvements for the Measurements panel added in https://github.com/nextstrain/auspice/pull/1452.

TODOs

Prioritized (ordered from highest to lowest priority)

• [x] #1814
• [x] #1815
• [ ] #1819
• [ ] #1816
• [ ] #1817

Not prioritized

• [ ] #1818
• [ ] #1820
• [ ] #1821
• [ ] #1822
• [ ] #1823
• [ ] #1824
• [ ] #1825
• [ ] #1826
• [ ] #1828

Completed

• [x] Order subplots by custom order from JSON (addressed in https://github.com/nextstrain/auspice/pull/1452)
• [x] Update Card title to handle long titles (see PR comment)
• [x] Add title to hover panel (color-by value? row name? - will need to inject color-by value since it is not included in the measurement's data)
• [x] Display "0" titer values in hover panel
• [x] Add color to row title if available (e.g. reference strains can have matching tree color-by)
• [x] Group raw measurements by color for y-axis jitter
• [x] multiple threshold lines (requested by @rneher)
• [x] sticky x-axis (requested by CDC)

[huddlej]

Playing with the measurements panel more today, I had a few comments/thoughts to add to the list above.

1. +1 for handling the long card titles. In the example community build, I can only get the tree and measurements to appear side-by-side when I close the control panel.

Control panel open:

Control panel closed:

2. In grid mode, there is no legend for measurement panel colors. It would be nice if the tooltip for grouped mean +/- SD mode also displayed the value associated with the color-by.

3. What do colors mean in measurements panel when coloring the tree by a continuous attribute like number of epitope mutations? Related to the point above about tooltips, it would be helpful to include the group info in the tooltips for the mean +/- SD mode.

4. In raw measurements display, it would be nice if the measurements could be grouped by color-by group like mean +/- SD, to minimize obscuring data points across groups. This would be aligned with how seaborn produces swarmplots with hues, as in this example from the original multiplot proposal issue:

5. Strains appear to have no value for log2 titer when plotted at log2 normalized titer of 0.0. Are the log2 values for these pairs actually 0.0 (and the issue is with the tooltip showing missing data) or are they missing (and the panel is somehow plotting them at 0.0 by default)?

6. Switching between collections resets the “Show overall mean +/- SD” and “Show measurements threshold” toggles and the “Group By” dropdown but does not reset the “Measurements display”. I was surprised to find that the overall mean toggle was not a setting for the whole panel instead of the collection being visualized. I'm not sure how we can communicate this subtle distinction in toggles in the UI though.

[jameshadfield]

What do colors mean in measurements panel when coloring the tree by a continuous attribute like number of epitope mutations?

Auspice groups continuous variables into a finite number of bins -- these are exposed in the legend, and hovering over each entry will highlight the matching tips. Bonus: this is user customisable. The measurements panel uses this same grouping. I agree that it would be nice to add this info to the on-hover tip box.

[jameshadfield]

Added to main list of TODOs

---

Some more improvements / experiments (via #1557)

• [ ] Link hover actions to tip radii changes in the tree
• [ ] Look into colouring categories by their underlying reference sera colours in the tree
• [ ] SVG text splitting of y-axis labels. See also this coment by @huddlej

[joverlee521]

Added to main list of TODOs

---

Another improvement / experiment idea from @trvrb via Slack:

General feature I've wanted for the Measurements panel is to be able to click on a row at the y-axis (maybe with a little gear or spotlight following the name) that would then color the tree by this row's measurements.

[huddlej]

Added to main list of TODOs

---

From @rneher:

In the measurement panel, some limits on the axis might be useful as well. We currently have -6 to 14 in some views (which is a million fold titer difference):