twopartCIM {twopartqtl}R Documentation

Composite interval mapping for point-mass mixtures

Description

Conducts composite interval mapping for phenotypes distributed as a point-mass mixture following the same procedure as QTL Cartographer. Significant covariate markers are chosen through a forward selection procedure up to a user-specified number. Selected markers are used as covariates in interval mapping procedure. Covariate marker covariates within a fixed, user-specified window of the location being tested are dropped.

Usage

twopartCIM(cross, pheno.col = 1, n.marcovar = 3, window = 10, pm.value = 0, 
          threshold = 1, maxit = 4000, tol = 1e-04, verbose = FALSE, 
          imp.method = c("imp", "argmax"), error.prob = 1e-04, 
          map.function = c("haldane", "kosambi", "c-v", "morgan"), use.log = FALSE, 
          n.perm)

Arguments

cross An object of class cross. See read.cross for details.
pheno.col Column number in the phenotype matrix to be used as the phenotype. Only one column can be analyzed at a time.
n.marcovar Number of marker covariates to use
window Window size, in cM
pm.value Value of the point-mass observations
threshold Significance threshold (p-value) for retaining covariates
maxit Maximum number of iterations
tol Tolerance value for determining convergence
verbose In the case n.perm is specified, displays information about the progress of the permutation tests.
imp.method Method used to impute any missing marker genotype data. See fill.geno for details.
error.prob Genotyping error probability assumed when imputing the missing marker genotype data.
map.function Map function used when imputing the missing marker genotype data.
use.log If TRUE phenotype values not in the point-mass are log transformed.
n.perm If specified, a permutation test is performed rather than an analysis of the observed data. This argument defines the number of permutation replicates.

Details

Missing marker genotype data are first imputed via fill.geno according to the specified imp.method.

Covariate markers are then identified through a forward selection process as described in Taylor and Pollard (2009). The most significant marker is retained at each step until n.marcovar markers are identified. If threshold is less than 1, a marker will be used as a covariate only if it is significant (i.e., p-value < threshold). Significant markers up to n.marcovar markers will be retained.

Selected markers are then used as covariates in interval mapping procedure. Covariate marker covariates within a fixed, user-specified window of the location being tested are dropped. The number of covariate markers is adjusted if necessary to be at least twice the number of continuous observations.

Value

The function returns an object of the same form as the function scanone:
If n.perm is missing, the function returns the scan results as a data.frame with three columns: chromosome, position, LOD score. Attributes indicate the names and positions of the chosen marker covariates.
If n.perm > 0, the function results the results of a permutation test: a vector giving the genome-wide maximum LOD score in each of the permutations.

Author(s)

Sandra L. Taylor, sltaylor@ucdavis.edu

References

Taylor, S.L. and K.S. Pollard 20XX. Composite interval mapping to identify quantitative trait loci for point-mass mixture phenotypes. Genetics Research, XX, xxx–xxx

See Also

cim

Examples


# Simulate backcross experiment
Map <- sim.map(c(100), n.mar=c(22), include.x=FALSE, eq.spacing=TRUE)
num.ind <- 200
QTL.info <- rbind(c(1,30,0.6,-0.02))
sim <- sim.cross.exp(map=Map, model=QTL.info, n.ind=num.ind, type="bc",
    keep.qtlgeno=TRUE, map.function="haldane", dist="pointmass", params=c(0.5,6,0.6))
sim <- calc.genoprob(sim, step=1)

out <- twopartCIM(sim, pheno.col=1, n.marcovar=3, pm.value=0)

# Permutation tests
## Not run: out.perm <- twopartCIM(sim, pheno.col=1, n.marcovar=3, pm.value=0, n.perm=1000)

[Package twopartqtl version 1.0 Index]