Learn R Programming

phyloland (version 1.0)

PLD_interface: Model parameters and internal locations estimations

Description

Function that estimates model parameters, genealogies and internal locations through Bayesian Markov chain Monte Carlo (MCMC) algorithm.

Usage

PLD_interface(fileTREES, fileDATA, num_step = 1e+05, freq = 100, 
burnin = 0, ess_lim = 100, sigma = NA, lambda = NA, tau = NA, 
num_step_sigma = 1, num_step_lambda = 1, num_step_tau = 1, 
id_filena = NA, pattern_trees_likelihood = "treeLikelihood", 
names_locations = NA)

Arguments

fileTREES
a character string that specifies the name of the file containing the phylogenetic trees. A Nexus file (e.g. BEAST output) with tips names (no space characters allowed), the phylogenetic trees and the trees likelihood. If only one tree is contained in th
fileDATA
a character string that specifies the name of the file containing the tips locations : a text file with 3 columns (separated by tabs). The first one with the tips names (the same as in fileTrees), the second one with the location latitudes an
num_step
a strictly positive integer specifying the length of the Markov Chain (the number of MCMC steps). Suggested values : 50 000 - 100 000.
freq
a strictly positive integer specifying how often the sampled model parameters and the locations are saved. Suggested values : 50 - 100.
burnin
a strictly positive integer specifying the number of trees dropped from the file fileTREES. Suggested value : 10% of the total number of trees, 0 in case of only one tree.
ess_lim
a strictly positive integer specifying the Effective Sample Size. The MCMC stops when the ess of all parameters reach this value.
sigma
a vector of two elements, strictly positive values specifying the standard deviation of the normal distribution for dispersal steps in each dimension. Default is NA: the 2 sigma parameters are estimated in the MCMC. Alternatively, if values are specified
lambda
a strictly positive value specifying the probability of migrating to an occupied location. Default is NA: the lambda parameter is estimated in the MCMC. Alternatively, if a value is specified the lambda parameter is not estimated by the MCMC. A lambda val
tau
a strictly positive value specifying the overall dispersal rate. Default is NA: the tau parameter is estimated in the MCMC. Alternatively, if a value specified the tau parameter is not estimated by the MCMC.
num_step_sigma
a strictly positive integer specifying how many times the sigma parameters are sampled at each MCMC step.
num_step_lambda
a strictly positive integer specifying how many times the lambda parameter is sampled at each MCMC step.
num_step_tau
a strictly positive integer specifying how many times the tau parameter is sampled at each MCMC step.
id_filena
an optional character string specifying the ID for the output files.
pattern_trees_likelihood
a character string specifying the text pattern to retrieve the trees likelihood in fileTREES.
names_locations
a vector of character strings that specifies the location names, in the same order as in fileDATA.

Value

  • an object of class "phyloland" with the following components:
  • treesan object of class "multiPhylo" or "phylo" (package ape) containing the trees sampled in the MCMC. If the ess_lim is not reached, the number of trees is num_step/freq.
  • locationsa matrix containing the internal locations sampled in the MCMC and the tips locations (columns) for each sampled tree(rows).
  • tipsa vector of character strings containing the tips names (from fileTREES and fileDATA).
  • spacea matrix containing the unique locations from fileDATA.
  • mcmca list containing the posterior distributions of each parameters.
  • sigma_limita vector indicating the threshold for sigma below which limited dispersal occurs.

Details

MCMC runs stop when the num_step is reached or when the ess of all estimated parameters equal ess_lim.

The respective ess for estimating convergence of the chains regarding each parameter is calculated by monitoring the autocorrelation between successive sampled states in the MCMC.