An MCMC sampler for loci using precomputed dispersion matrices, various priors, and a pre-selected set of variables. For use with F2 intercross design.
Using precomputed dispersion matrices, various priors, and a pre-selected set of variables, one locus is removed, all other loci are examined to determine the effect of each on the posterior. One locus is sampled. The process is repeated until each locus has been removed and a new one sampled in its place (possibly the same one that was removed is sampled).
swapf2(varcov, invars, rparm, nreps, ana.obj, locs,
locs.prior, combo.prior, tol = 1e-10)A list with components:
A 2k by k by nreps array of indexes of variables sampled in
each of the nreps iterations. Models using less than 2k variables
configs[,i,j] will contain one or more zeroes in the last
position(s)
A vector of length k*nreps with the
posteriors of the models sampled.
A 2k by k by nreps matrix of the regression
coefficients. Models using less than 2k variables
configs[,i,j] will contain one or more zeroes in the last
position(s)
The call to swapf2
The k*nreps posterior probabilities of the k-1 gene
models.
The k*nreps marginal posteriors for all k gene
models that could be formed using the current k-1 gene model)
A vector with length(locs) elements. At
each step, the posterior associated with each candidate locus is
added to an element of this vector. After all steps are finished,
the result is normalized to sum to one. This turns out to be an
exceedingly stable estimate of the relative marginal posterior.
A vector with 2*length(locs) elements. At
each step, the product of each posterior times the coefficient
associated with a candidate variable is
added to an element of this vector. After all steps are finished,
the result is normalized by the total marginal posterior. This turns
out to be an rather stable estimate of the marginal (over all
models) posterior mean of the regression coefficients.
The result of make.varcov . The columns of
varcov$var.x must alternate 'additive' and 'dominance' terms.
The 'ridge' parameters for the independent variables - larger values imply more shrinkage or a more concentrated prior for the regresion coefficients.
How many cycles of MCMC to perform
A object produced by make.analysis.obj
A vector of variable indexes. This determines which
variables to start in the model. If both additive and
dominance terms are to be used, they should occupy adjacent
locations in invars. The variable(s) associated with the
first locus is (are) immediately removed, serving only as
placeholder(s). If there are k loci associated with the variables,
then all subsequent models have k loci, although the nuimber of
variables may vary according to the selection of one or both of the
'additive' or 'dominance' terms.
The pairs of columns of varcov$var.x to use. The default
uses all of them.
Vector whose elements are the prior masses to associate with each locus. Typically, these sum to one, but sometimes they might each be set to one (as in computing lod scores). The default value sets them all to 1.0.
The prior probability for each term or combination of terms for the phenotypic effect at a locus. Typically, there will be three of these - one for the 'additive' term (linear in number of alleles from one parent strain), the 'dominance' term (quadratic in allele number), or both terms. The default sets them all to 1/3.
Used in forming QR decomposition. Let it be.
Charles C. Berry cberry@ucsd.edu
A call to swapf2 is used to obtain the results. This function
is really just a wrapper.
Berry C.C. (1998) Computationally Efficient Bayesian QTL Mapping in Experimental Crosses. ASA Proceedings of the Biometrics Section, 164-169.
swapbc1