get_ML_states_from_relprobs: Extract the ML states at each node, from a table of relative...
In BioGeoBEARS: BioGeography with Bayesian (and Likelihood) Evolutionary Analysis in R Scripts

Description Usage Arguments Details Value Note Author(s) References See Also Examples

View source: R/BioGeoBEARS_simulate_v1.R

Given a table with the rows representing nodes, and the columns representing the relative probabilities of each state, this function finds the ML (maximum likelihood) state(s) for each node.

1 2	get_ML_states_from_relprobs(relprobs, statenames, returnwhat = "states", if_ties = "takefirst")

`relprobs`	A numeric matrix of relative probabilities
`statenames`	The names of the states/geographic ranges (e.g., A, AB, CDE, ABD, etc...)
`returnwhat`	If "indices", return the 0-based indices of the states. If "states", return the name of the state, based on statenames.
`if_ties`	What to do with ties. Currently, the only option is to take the first (this will be shown in e.g. a pie chart, of course).

If possible, the input matrix should be the actual ML estimate of the state probabilities at each node, rather than just the scaled conditional likelihoods at each node. The latter reflect only the tips-down information, whereas the former (the marginal ancestral state reconstruction) uses all of the information, and the probabilities of the states at the root and in the outgroup(s) can influence the estimates in the ingroups. This would not likely be particularly important in a pure continuous-time model, but in a model with cladogenesis it could matter quite a bit.

See http://blog.phytools.org/2013/03/marginal-ancestral-state-reconstruction.html for more discussion of marginal ancestral state reconstructions, versus mere scaled conditional likelihoods.

Revell and other sources (Felsenstein (2004)) advocate the "re-rooting" method for obtaining the marginal ancestral state reconstructions; however, re-rooting requires a time-reversible model and a tree with no root. In biogeography we have a non-reversible model, and typically a time-scaled chronogram. However, the same result can be obtained by modifying the scaled conditional likelihoods obtained from a downpass from the tips, via an doing an up-pass from the root scaled conditional likelihoods, being careful to transfer probabilities via the time-forward version of the Q-matrix and cladogenesis/speciation matrix.

Note: further notes as this is implemented (required!)

ML_states or ML_states_indices, depending on returnwhat.

Go BEARS!

Nicholas J. Matzke matzke@berkeley.edu

http://phylo.wikidot.com/matzke-2013-international-biogeography-society-poster http://blog.phytools.org/2013/03/marginal-ancestral-state-reconstruction.html https://code.google.com/p/lagrange/

Felsenstein2004

Matzke_2012_IBS

get_ML_state_indices

testval=1

Loading required package: rexpokit
Loading required package: cladoRcpp
Loading required package: ape
Loading required package: phylobase

Attaching package: 'phylobase'

The following object is masked from 'package:ape':

    edges