fitStahl: Fit Stahl model

In kbroman/xoi: Tools for Analyzing Crossover Interference

Fit Stahl model

Description

Fit the Stahl model for crossover interference to data on crossover locations.

Usage

fitStahl(
  xoloc,
  chrlen = NULL,
  nu = c(1, 20),
  p = 0.02,
  max.conv = 25,
  integr.tol = 0.00000001,
  max.subd = 1000,
  min.subd = 10,
  verbose = TRUE,
  ...
)

Arguments

xoloc	A list of crossover locations (in cM), each component being a vector of locations for a different meiotic product.
chrlen	Chromosome length (in cM), either of length 1 or the same length as xoloc.
nu	Interference parameter (\nu). This should be a pair of values to be used as endpoints to first do a 1-dimensional optimization with p=0.
p	Starting value for the proportion of crossovers from the no interference pathway, for the 2-dimensional optimization.
max.conv	Maximum limit for summation in the convolutions to get inter-crossover distance distribution from the inter-chiasma distance distributions. This should be greater than the maximum number of chiasmata on the 4-strand bundle.
integr.tol	Tolerance for convergence of numerical integration.
max.subd	Maximum number of subdivisions in numerical integration.
min.subd	Minimum number of subdivisions in numerical integration.
verbose	If TRUE, print tracing information. If "..." includes control, this is ignored.
...	Further arguments sent to stats::optim().

Details

See Housworth and Stahl (2003) and Broman and Weber (2000) for details of the method.

We first use stats::optimize() to find the MLE with the contraint p=0, followed by use of stats::optim() to do a 2-dimensional optimization for the MLEs of the pair.

Value

A vector with the estimates of \nu (interference parameter) and p (proportion of crossovers coming from the no interference pathway), the maximized log likelihood, the estimate of nu with p constrained to be 0, the maximized log likelihood in this case, and the log likelihood ratio for comparing the model with p allowed to vary freely versus contrained to be 0. (Note that it's the natural log of the likelihood ratio, and not twice that.)

Author(s)

Karl W Broman, [e-mail]

References

Housworth, E. A. and Stahl, F. W. (2003) Crossover interference in humans. Am. J. Hum. Genet. 73, 188–197.

Broman, K. W. and Weber, J. L. (2000) Characterization of human crossover interference. Am. J. Hum. Genet. 66, 1911–1926.

See Also

fitGamma(), stahlLoglik(), simStahl()

Examples

data(bssbsb)


xoloc <- find.breaks(bssbsb, chr=1)
L <- attr(xoloc, "L")

# get MLE (limiting maximum iterations to 10, just for speed in this example)
## Not run: mle <- fitStahl(xoloc, L, nu=c(9, 12), control=list(maxit=10))

kbroman/xoi documentation built on May 1, 2023, 9:35 p.m.