Rsampletrees (version 1.0.3)

estimateHap: Estimate and write haplotype probabilities and initial values to files

Description

This function is only used when the data type is `g' (genotype). Estimate the two-locus haplotype probabilities from the genotype data. Optionally set initial haplotype configurations and a list of likely haplotypes for the run of sampletrees. This function requires the R package "haplo.stats" for estimating the haplotype frequencies from the genotype data.

Usage

estimateHap(args, HaploFreqFile, InitialHaplos = TRUE, 
InitialHaploFile = "initialhaps", HaploList = TRUE, 
HaploListFile = "initialhaplist", tol = 1e-05)

Arguments

args

An object of class `pars' with the sampletrees settings

HaploFreqFile

The name of the output file for the two-locus haplotype probability estimates

InitialHaplos

If TRUE, sample the initial haplotype configuration using the posterior probabilities of each configuration for each individual estimated using haplo.em() (Default=TRUE)

InitialHaploFile

File name for the initial haplotype configurations

HaploList

If TRUE, create a list of likely haplotypes for a run of sampletrees with genotype data (Default=TRUE). This list will include haplotypes estimated to have a probability greater than 'tol'

HaploListFile

File name for the haplotype list

tol

Haplotypes with estimated probability greater than this value will be included in the list of likely haplotypes

Value

This function writes the estimated haplotype data to the specified files and returns

args

An object of class `pars' with the haplotype options set to those specified by the call to this function

Details

This function is only used when the data type is genotype (`g').

The two-locus haplotype probabilities are estimated using the haplo.em() function in the haplo.stats package. This package uses an EM approach that has been adapted to handle estimation of haplotype probabilities when the haplotypes are made up of many loci. The haplotype probabilities are estimated for haplotypes containing all loci. The probability for a given two-locus haplotype is then computed by summing up probabilities for the full haplotypes having the given two-locus haplotype (possible haplotypes are 00, 01, 10 or 11). These probabilities are computed for all adjacent pairs of loci.

When using genotype data, it is recommended that a list of likely or known haplotypes and an initial configuration be provided to sampletrees in order to improve MCMC mixing. These can optionally be initialized using the output from haplo.em() if HaploList and InitialHaplos are set to TRUE. The list of likely haplotypes will contain those haplotypes that have probability estimated to be above a threshold (set by `tol'). The initial haplotype configuration for all individuals is initialized by sampling a configuration based on the estimated posterior probabilities of each haplotype configuration for each individual.

References

Burkett KM, McNeney B, Graham J. Sampletrees and Rsampletrees: sampling gene genealogies conditional on SNP genotype data. Bioinformatics. 32:1580-2, 2016

Examples

# NOT RUN {
datname=paste(path.package("Rsampletrees"),"/extdata/geno_Theta8_Rho8.txt",sep="")
locname=paste(path.package("Rsampletrees"),"/extdata/locations_Theta8_Rho8.txt",sep="")
weightname=paste(path.package("Rsampletrees"),"/extdata/weights-g.txt",sep="")

runpars=newArgs(DataFile=datname, DataType="g", LocationFile=locname, WeightFile=weightname,
	    RunName="Test-g",FocalPoint=10000)

# This will create the three files"EM-hapfreqs", "EM-initial.dat", "EM-known_haplotypes"
runpars=estimateHap(runpars,"EM-hapfreqs",InitialHaploFile="EM-initial.dat",
HaploListFile="EM-known_haplotypes")

# }