haplo.scan: Search for a trait-locus by sliding a fixed-width window over...

In haplo.stats: Statistical Analysis of Haplotypes with Traits and Covariates when Linkage Phase is Ambiguous

Search for a trait-locus by sliding a fixed-width window over each marker locus and scanning all possible haplotype lengths within the window

Description

Search for haplotypes that have the strongest association with a binary trait (typically case/control status) by sliding a fixed-width window over each marker locus and scanning all possible haplotype lengths within the window. For each haplotype length, a score statistic is computed to compare the set of haplotypes with a given length between cases versus controls. The locus-specific score statistic is the maximum score statistic calculated on loci containing that locus. The maximum score statistic over all haplotype lengths within all possible windows is used for a global test for association. Permutations of the trait are used to compute p-values.

Usage

haplo.scan(y, geno, width=4, miss.val=c(0, NA),
          em.control=haplo.em.control(),
          sim.control=score.sim.control())

haplo.scan.obs(y, em.obj, width)

haplo.scan.sim(y.reord, save.lst, nloci)

Arguments

y	Vector of binary trait values, must be 1 for cases and 0 for controls.
y.reord	Same as y, except the order is permuted
geno	Matrix of alleles, such that each locus has a pair of adjacent columns of alleles, and the order of columns corresponds to the order of loci on a chromosome. If there are K loci, then ncol(geno) = 2*K. Rows represent alleles for each subject.
width	Width of sliding the window
miss.val	Vector of codes for missing values of alleles
em.control	A list of control parameters to determine how to perform the EM algorithm for estimating haplotype frequencies when phase is unknown. The list is created by the function haplo.em.control - see this function for more details.
sim.control	A list of control parameters to determine how simulations are performed for simulated p-values. The list is created by the function score.sim.control and the default values of this function can be changed as desired. See score.sim.control for details.
em.obj	Object returned from haplo.em, performed on geno
save.lst	Information on haplotypes needed for haplo.scan.sim, already calculated in haplo.scan
nloci	number of markers

Details

Search for a region for which the haplotypes have the strongest association with a binary trait by sliding a window of fixed width over each marker locus, and considering all haplotype lengths within each window. To acount for unknown linkage phase, the function haplo.em is called prior to scanning, to create a list of haplotype pairs and posterior probabilities. To illustrate the scanning, consider a 10-locus dataset. When placing a window of width 3 over locus 5, the possible haplotype lengths that contain locus 5 are three (loci 3-4-5, 4-5-6, and 5-6-7), two (loci 4-5, and 5-6) and one (locus 5). For each of these loci subsets a score statistic is computed, which is based on the difference between the mean vector of haplotype counts for cases and that for controls. The maximum of these score statistics, over all possible haplotype lengths within a window, is the locus-specific test statistic. The global test statistic is the maximum over all computed score statistics. To compute p-values, the case/control status is randomly permuted. Simulations are performed until precision criteria are met for all p-values; the criteria are controlled by score.sim.control. See the note for long run times.

Value

A list that has class haplo.scan, which contains the following items:

call	The call to haplo.scan
scan.df	A data frame containing the maximum test statistic for each window around each locus, and its simulated p-value.
max.loc	The loci (locus) which contain(s) the maximum observed test statistic over all haplotype lengths and all windows.
globalp	A p-value for the significance of the global maximum statistic.
nsim	Number of simulations performed

Note

For datasets with many estimated haplotypes, the run-time can be very long.

References

Cheng et al-1

Cheng R, Ma JZ, Wright FA, Lin S, Gau X, Wang D, Elston RC, Li MD. "Nonparametric disequilibrium mapping of functional sites using haplotypes of multiple tightly linked single-nucleotide polymorphism markers". Genetics 164 (2003):1175-1187.

Cheng et al-2

Cheng R, Ma JZ, Elston RC, Li MD. "Fine Mapping Functional Sites or Regions from Case-Control Data Using Haplotypes of Multiple Linked SNPs." Annals of Human Genetics 69 (2005): 102-112.

See Also

haplo.em, haplo.em.control, score.sim.control

Examples

  # create a random genotype matrix with 10 loci, 50 cases, 50 controls
  set.seed(1)
  tmp <- ifelse(runif(2000)>.3, 1, 2)
  geno <- matrix(tmp, ncol=20)
  y <- rep(c(0,1),c(50,50))

  # search 10-locus region, typically don't limit the number of
  # simulations, but run time can get long with many simulations

  scan.obj <- haplo.scan(y, geno, width=3,
                sim.control = score.sim.control(min.sim=10, max.sim=20))

  print(scan.obj)

haplo.stats documentation built on May 29, 2024, 9:53 a.m.