sim_jive: Simulate JIVE process
In bite: Bayesian Integrative Models of Trait Evolution

Description Usage Arguments Details Value Author(s) Examples

Generate random values of trait mean and variance simulated under a JIVE process along a phylogenetic tree

sim_jive(phy, map = NULL, models = list(mean = c("BM"), logvar =
  c("OU")), pars = list(mean = c(root = 0, sigma_sq = 0.1), logvar =
  c(root = 2, theta = 1, sigma_sq = 0.1, alpha = 1)), sampling = c(1, 7),
  bounds = list(c(-Inf, Inf), c(-Inf, Inf)), var.f = NULL)

`phy`	Phylogenetic tree
`map`	list containing the mapping of regimes over each edge (see details)
`models`	list giving model specification for the simulation of trait evolution. Supported models are c("OU", "BM", "WN"). The user can also specify if the assumptions of the model should be relaxed (see details)
`pars`	list giving parameters used for the simulation of trait evolution (see details). Name of parameters must be explecitly entered (see details)
`sampling`	vector of size 2 giving the min and max number of individual per species
`bounds`	list giving traits bounds
`var.f`	alternative function to model intraspecific variation of the form: function(n, pars) with n, number of individuals and pars, parameters of the model

map : the list must be ordered in the same order than phy$edge. Each element represents an edge and contains a vector indicating the time spent under each regime in the branch. The name of the regimes must appear on the map models : trait evolution can be simulated using Ornstein-Uhlenbeck (OU), Brownian Motion (BM) or White Noise (WN) processes. Multiple regimes can be defined for both models and will apply on thetas: c("OU", "theta"), sigmas: c("OU", "sigma") or alphas: c("OU", "alpha") for OU and on sigmas only for WN: c("WN", "sigma") and BM: c("BM", "sigma"). While using the OU model, the user can also relax the stationarity of the root: c("OU", "root") and relax several assumptions at the same time c("OU", "root", "theta") pars : list containing parameters depending on the chosen model. Elements of that lists must be vectors of size 1 or n, with n = number of regimes in the map. Each element of pars must be named with the corresponding parameter abbreviation. Parameters used in the different models:

White Noise model (WN):

root: root value
sigma_sq: evolutionary rate, n regimes if "sigma" is specified in models

Brownian Motion model (BM):

root: root value
sigma_sq: evolutionary rate, n regimes if "sigma" is specified in models

Ornstein Uhlenbeck model (OU):

root: root value. Only used if "root" is specified in models
sigma_sq: evolutionary rate, n regimes if "sigma" is specified in models
theta: optimal value, n regimes if "theta" is specified in models
alpha: strength of selection, n regimes if "alpha" is specified in models

A list of length two containing a numeric matrix named "evo_traits" giving simulated traits representing intraspecific variation and a data.frame called "obs" containing species name in the fisrt column and individual observation of the trait in the second.

Theo Gaboriau

library(phytools)
phy <- pbtree(n = 50)
Q <- cbind(c(-.002, .002), c(.002, -.002))
phy <- sim.history(phy, Q = Q)
# MBM and VOU
jive_phy <- sim_jive(phy = phy, map = phy$maps)

# MWN + sigma and VOU + theta + root + alpha
jive_phy <- sim_jive(phy = phy, map = phy$maps, 
   models = list(mean= c("WN", "sigma"), logvar = c("OU")),
   pars = list(mean = c(root = 0, sigma_sq1 = 0.1, sigma_sq2 = 0.5), 
              logvar = c(root = 10, theta1 = 5, theta2 = 10,
                         sigma_sq = 0.1, alpha1 = 0.2, alpha2 = 0.8))
 )


# With a different model of intraspecific variation:
unif.f <- function(n, pars){
 runif(n, pars[1] - exp(pars[2])/2, pars[1] + exp(pars[2])/2)
}

unif_phy <- sim_jive(phy = phy, map = phy$maps, models = list(mid=c("BM"), logrange=c("OU")),
 pars = list(mid = c(root = 0, sigma_sq = 0.1),
             logrange = c(root = 2, theta = 1, sigma_sq = 0.1, alpha = 1)),
 var.f = unif.f)