simulate_rou_model {castor}R Documentation

Simulate a reflected Ornstein-Uhlenbeck model for continuous trait evolution.

Description

Given a rooted phylogenetic tree and a reflected Ornstein-Uhlenbeck (ROU) model for the evolution of a continuous (numeric) trait, simulate random outcomes of the model on all nodes and/or tips of the tree. The ROU process is similar to the Ornstein-Uhlenbeck process (see simulate_ou_model), with the difference that the ROU process cannot fall below a certain value (its "reflection point"), which (in this implementation) is also its deterministic equilibrium point (Hu et al. 2015). The function traverses nodes from root to tips and randomly assigns a state to each node or tip based on its parent's previously assigned state and the specified model parameters. The generated states have joint distributions consistent with the ROU model. Optionally, multiple independent simulations can be performed using the same model.

Usage

simulate_rou_model(tree, reflection_point, spread, decay_rate,
                   include_tips=TRUE, include_nodes=TRUE, 
                   Nsimulations=1, drop_dims=TRUE)

Arguments

tree

A rooted tree of class "phylo". The root is assumed to be the unique node with no incoming edge.

reflection_point

Numeric. The reflection point of the ROU model. In castor, this also happens to be the deterministic equilibrium of the ROU process (i.e. if the decay rate were infinite). For example, if a trait can only be positive (but arbitrarily small), then reflection_point may be set to 0.

spread

Numeric. The stationary standard deviation of the corresponding unreflected OU process.

decay_rate

Numeric. Exponential decay rate (stabilization rate) of the ROU process (in units 1/edge_length_units).

include_tips

Include random states for the tips. If FALSE, no states will be returned for tips.

include_nodes

Include random states for the nodes. If FALSE, no states will be returned for nodes.

Nsimulations

Number of random independent simulations to perform. For each node and/or tip, there will be Nsimulations random states generated.

drop_dims

Logical, specifying whether the returned tip_states and node_states (see below) should be vectors, if Nsimulations==1. If drop_dims==FALSE, then tip_states and tip_nodes will always be 2D matrices.

Details

For each simulation, the state of the root is picked randomly from the stationary distribution of the ROU model, i.e. from a one-sided normal distribution with mode = reflection_point and standard deviation = stationary_std.

If tree$edge.length is missing, each edge in the tree is assumed to have length 1. The tree may include multi-furcations (i.e. nodes with more than 2 children) as well as mono-furcations (i.e. nodes with only one child). The asymptotic time complexity of this function is O(Nedges*Nsimulations), where Nedges is the number of edges in the tree.

Value

A list with the following elements:

tip_states

Either NULL (if include_tips==FALSE), or a 2D numeric matrix of size Nsimulations x Ntips, where Ntips is the number of tips in the tree. The [r,c]-th entry of this matrix will be the state of tip c generated by the r-th simulation. If drop_dims==TRUE and Nsimulations==1, then tip_states will be a vector.

node_states

Either NULL (if include_nodes==FALSE), or a 2D numeric matrix of size Nsimulations x Nnodes, where Nnodes is the number of nodes in the tree. The [r,c]-th entry of this matrix will be the state of node c generated by the r-th simulation. If drop_dims==TRUE and Nsimulations==1, then node_states will be a vector.

Author(s)

Stilianos Louca

References

Y. Hu, C. Lee, M. H. Lee, J. Song (2015). Parameter estimation for reflected Ornstein-Uhlenbeck processes with discrete observations. Statistical Inference for Stochastic Processes. 18:279-291.

See Also

simulate_ou_model, simulate_bm_model, simulate_mk_model

Examples

# generate a random tree
tree = generate_random_tree(list(birth_rate_intercept=1),max_tips=10000)$tree

# simulate evolution of a continuous trait whose value is always >=1
tip_states = simulate_rou_model(tree, reflection_point=1, spread=2, decay_rate=0.1)$tip_states

# plot histogram of simulated tip states
hist(tip_states, breaks=20, xlab="state", main="Trait probability distribution", prob=TRUE)

[Package castor version 1.7.0 Index]