fitqtl: Stepwise backward elimination and anova comparison.
In byandell/qtlbim: QTL Bayesian Interval Mapping

Description Usage Arguments Details Value Author(s) References See Also Examples

These functions mimic step and anova but have reduced functionality. They are not truly methods, but can help study qtl model fits.

step.fitqtl(cross, qtl, pheno.col = 1, arch, cutoff = 0.05,
  trace = 1, steps = 100)
## S3 method for class 'step.fitqtl'
anova(object, object2, ...)

`cross`	Object of class `cross`.
`qtl`	Object of class `qtl`, as output of `makeqtl`.
`pheno.col`	Column of phenotype (numeric).
`arch`	Object of class `qb.arch` from `qb.arch`.
`cutoff`	Significance cutoff for dropping terms.
`trace`	If positive, information is printed during the run. Values 1, 2, 3 give gradually more detailed information.
`steps`	Maximum number of steps to be considered.
`object`	Object of class `step.fitqtl` from `step.fitqtl`.
`object2`	Object of class `step.fitqtl` from `step.fitqtl`.
`...`	Currently not used.

step.fitqtl is analogous to step applied to analysis with fitqtl. anova.step.fitqtl is an S3 method for anova. anova.step.fitqtl with one argument calls summary.fitqtl; with two arguments it attempts to conduct a general F comparison of anova fits.

step.fitqtl returns an object of class step.fitqtl with

`fit`	Object of class `fitqtl`.
`arch`	Object of class `qb.arch`.

Brian S. Yandell, yandell@stat.wisc.edu

http://www.qtlbim.org

qb.arch, fitqtl, summary.fitqtl, makeqtl

cross <- qtl::sim.geno(cross, n.draws = 8, step = 2, error = 0.01)
qtl <- qtl::makeqtl(cross, chr = c(1,1,2,3), pos = c(15,45,12,15))
cross.step <- step.fitqtl(cross, qtl, pheno.col = 3, arch = cross.arch)
anova(cross.step)
cross.step$arch