How to make state transition distribution plots (in 3 easy steps)

20 January 2021 by Remco Bouckaert

This is a how-to on making plots like this:

Introductions into NZ

Counting the number of introductions from individual places

Introductions into NZ

Counting the number of introductions from all other places

Lineages through time

Lineages through time being at a particular location

This involves 3 steps

  • Generate tree file with BEAST
  • Run StateTransitionCounter
  • Plot StateTransitionCounter results in bar graph

1. Generate tree file with BEAST

Set up an analysis that stores location/deme information in a tree file, e.g., like a DTA analysis, but Mascot and BDMM analyses or anything else that stores states as meta data should work as well. The tree file should have metadata on the nodes of the form tag=value pairs where tag can be location and value one of the demes under consideration.

2. Run StateTransitionCounter

StateTransitionCounter is a tool in Babel v0.3.0 (use the package manager to install, or build Babel from source, and install by hand). Once it is installed, you can run it via the File/Launch Apps menu in BEAUti, which then shows a dialog like so:

StateTransitionCounter dialog

or on a terminal with the applauncer, e.g. like so:

/path/to/beast/bin/applauncher StateTransitionCounter -tag location -in beast.trees

where /path/to/beast the path to where BEAST is installed on your computer.

StateTransitionCounter takes the following arguments:

  • in (TreeFile): source tree (set) file (required)
  • tag (String): label used to report trait (required)
  • out (OutFile): output file, or stdout if not specified (optional, default: [[none]])
  • burnin (Integer): percentage of trees to used as burn-in (and will be ignored) (optional, default: 10)
  • resolution (Integer): number of steps in lineages through time table (optional, default: 1000)

NB 1. On Windows with BEAST v2.6.2, this should be \path\to\BEAST\bat\applauncher.bar instead of /path/to/beast/bin/applauncher.

NB 2. Note that applauncher allows abbreviating its arguments, so the following should run the same:

/path/to/beast/bin/applauncher StateT -t location -i beast.trees

StateTransitionCounter counts the number of branches in a tree that have a certain state at the parent and another at the node. It produces a report like so:

Transition mean 95%Low 95%High
Asia=>Asia 162.16759156492785 156.0 170.0
Asia=>Europe 10.34850166481687 5.0 17.0
Asia=>New_Zealand 0.5349611542730299 0.0 2.0
Asia=>RestWorld 6.889012208657047 4.0 10.0
Europe=>Asia 0.7613762486126526 0.0 2.0
Europe=>Europe 124.22419533851276 112.0 135.0
Europe=>New_Zealand 6.407325194228635 3.0 11.0
Europe=>RestWorld 11.894561598224195 8.0 17.0

Histogram

Transition 0 1 2 3 4 5
Asia=>Asia 0 0 0 0 0 0
Asia=>Europe 0 0 0 0 8 28
Asia=>New_Zealand 552 225 115 9 0 0

Lineages through time

Transition 0.0 4.211E-4 8.422E-4 0.001263 0.001684 0.002105
Asia 2.0 2.0 2.0 2.0 2.0 2.0
Europe 0.0 0.0 0.0 0.0 0.0 0.0
New_Zealand 0.0 0.0 0.0 0.0 0.0 0.0
RestWorld 0.0 0.0 0.0 0.0 0.0 0.0

In the first table, the first columns represent the state transition with first the parent state and second the node state. The mean column is the mean number observed in the tree set, and 95%Low and 95%Hi are the 2.5% fraction and 97.5% fraction of the distribution (so, not necessarily the 95% HPD).

The second table is a histogram of transition counts, with the first row representing the number of counts and the first column representing the number of trees with zero transitions, next column 1, etc.

The third table represents a lineage through time plot, where the first row is the time and the numbers in following rows represent how many branches are in a particular state/location.

Create bar graph

The numbers from the table produced by StateTransitionCounter can be manipulated, for example in a spread sheet, and the rows and columns of interest selected from them.

transition 0 1 2 3 4 5 6 7 8 9 10 11
Asia=>New_Zealand 552 225 115 9                
Europe=>New_Zealand 0 0 23 131 187 180 130 100 74 49 25 2
RestWorld=>New_Zealand 311 25 39 66 109 143 136 50 14 8    

You can then export them as a graph (like the ones at the start of this page), or export to R and apply you favourite transformation on it.

NB To create a bar graph that contains transitions between groups of values, just open the tree file in a text editor and search/replace values. For example, the second graph at the top of the page is from the same tree set as the first, but with value location="Asia", location="RestWorld", and location="Europe" all replaced by location="RestOfWorld".