abctools (version 0.2-2)

pls.abc: ABC inference using PLS components computed from data.

Description

The function performs ABC inference using PLS transformed summary statistics.

Usage

pls.abc(obs, param, sumstats, obspar=NULL, abcmethod=abc,transfile = "Routput_test", 
bc=FALSE, err.only=TRUE, errfn=rsse,...)

Value

A list with the following components:post.samplean array of dimension nacc x npar x ndatasets giving the posterior sample for each observed dataset. Not returned if err.only=TRUE.errsimulation error (if obspar is supplied).

Details

The function uses a precomputed PLS transformation file to construct new summary statistics, being linear combinations of the original summary statistics. This is achieved using the transformer command line script supplied with the ABCtoolbox software package. Note that the precomputed transformation file can be created for example, with the find_pls.r R script supplied in ABCtoolbox. An example Routput_test for three components is supplied as a dataset. See the package documentation for more information on these two script files.

References

Blum, M. G. B, Nunes, M. A., Prangle, D. and Sisson, S. A. (2013) A comparative review of dimension reduction methods in approximate Bayesian computation. Stat. Sci. (to appear). Wegmann, D. et al. (2010) ABCtoolbox: A versatile toolkit for approximate Bayesian computations. BMC Bioinformatics 11, 116--123. Wegmann, D. et al. (2009) Efficient approximate Bayesian computation coupled With Markov chain Monte Carlo Without Likelihood. Genetics 182 (4), 1207--1218.

Examples

# example transformation file:

data(Routput_test)
write.table(Routput_test,file="Routput_test",quote=FALSE,row.names=FALSE,col.names=FALSE)

data(coal)

pls.abc(coal[1:3,3:7], coal[,1:2], coal[,3:7], transfile = "Routput_test",tol=.1,
method="rejection")