EnvironmentalDemogeneticsABC

Statistical models of coalescent on a graph

Coalescence in an Approximate Bayesian Computation framework

Model

Use askListOfParameters function to precise the model. The function will successively ask for a model, for the prior distribution of each parameter, and finally for values of hyperparameters. Then parameters values are drawn and stored with the model in a list object.

Coalescent simulations

Use simSpatialCoal to simulate spatial coalescence.

Required objects

The model list, a rasterStack of environment variables and a **spatial genetic dataset** :

Function

The function makes use of mclapply (package parallel) to perform parallel computing. This function relies on forking implementation, but Windows does not support forking. Furthermore, even on Linux, mclapply seems to behave quite weirdly. This will be settled later.

Results

simSpatialCoal function creates a repertory in the current directory, named SimulResults, where are written the simulated genetic values for each locus and individuals.

ABC analysis

Makes use of abc package functions.

Summary statistics

Use pca4abc function to computes summary statistics of simulated genetic dataset. Firstly, PC are computed according to the observed genetic data, then transformation is applied to all simulated genetic data. This function returns a list of three objects that can be used in abc package functions : a vector of observed summary statistics, a matrix of simulated summary statistics, and a matrix of parameters values used for simulations.

Cross validation

abc package can perform cross validation, using cv4abc

ABC

abc function of abc package performs a classic analysis

Forward models