Local Branching Index

[jameshadfield]

This adds LBI to nextstrain/auspice - code copied from nextflu/auspice. It runs when lbi is specified by augur as a color_option so it's backward compatible with all JSONs. Augur branch -
https://github.com/nextstrain/augur/tree/LBI-auspice - has this output for flu, and defines tau and time_window parameters.

Questions / observations (@rneher):

• The values are over [0, 1.0] but the colour scale is bounded by [0, 0.7] (same as nextflu)
• This takes around 18ms (2000 tip flu tree) to calculate, and does not use visibility (so filters / time selected doesn't affect it, and it's not recomputed when these things change). Is this OK, or should we aim to optimise things so it can be recomputed when the visibility changes?
• The colour scales look different between nextflu & nextstrain because the latter has (a) slightly different colour ramps and (b) picks values uniformly through the range -- this shows the same data with nexflu (top) vs nextstrain (below):
  

[rneher]

great! two quick suggestions:

• if you change the range to [0, 0.72] you get pretty numbers. the reason for the range [0,0.7] is purely aesthetic. if the range is [0, 1], then you rarely see red anywhere bc the peaks are internal nodes without circles.
• the calls to Math.exp could be the most costly bit. we could precompute node.ebl =Math.exp(-node.clock_length/LBI_tau). But I don't think we need to update this dynamically right now.

[jameshadfield]

Great - changing to 0.72 doesn't work as expected - i'll work on the legend sometime soon as there's a few little things there (0 isn't displayed, numbers could be better displayed...)

[rneher]

that's some messed up rounding. should be 0, 0.08, 0.16, 0.32, 0.4, etc

[jameshadfield]

for referenct - this is an example of a valid color_option in the meta JSON to generate LBI in auspice:

"lbi": {
 "menuItem": "local branching index",
 "tau": 0.4,
 "legendTitle": "local branching index",
 "key": "LBI",
 "timeWindow": 0.6
},