bim.model {bim}R Documentation

Bayesian model selection for number and pattern of QTL across genome


Posterior number and pattern of QTL, along with posterior/prior Bayes factor ratios.


bim.model( bim, cross, nqtl = 1, pattern=NULL, exact=FALSE,
  cutoff = 1 )
bim.nqtl( bim )
bim.pattern( bim, cross, nqtl = 1, pattern=NULL, exact=FALSE,
  cutoff = 1 )


bim object of class bim
cross corresponding object of class cross (extracted by bim.cross if not provided)
nqtl subset on number of QTL
pattern subset on chromosome pattern of QTL
exact subset on exact pattern or number of QTL if true
cutoff percent cutoff for inclusion in model selection


bim.model creates results from both bim.nqtl and bim.pattern.

bim.nqtl estimates posterior frequency of number of QTLs as the margine over all other model parameters. However, note that posterior may be influenced by prior, while Bayes factor is empirically less sensitive for QTL model selection. Bayes factors are ratios of bf=posterior/prior ratios.

bim.pattern shows at most 15 model patterns with at least cutoff % posterior are returned. Patterns are comma-separate list of chromosomes, with asterisk * for multiple QTL per chromosome. bim is first subsetted using subset.bim.


List with items nqtl and pattern, each containing:

posterior posterior for number of QTL
prior prior for number of QTL
bf rank-ordered posterior/prior ratios rescaled so bf[1] = 1
bfse approximate standard error for bf computed using binomial variance

In addition, there is an object param with values for nqtl, pattern, exact and cutoff.


Brian S. Yandell,


See Also



data( verngeo.bim )

bim.model( verngeo.bim )

[Package bim version 1.01-1 Index]