qtlmt (version 0.1-1)

mtcmimStep: MTCMIM model selection

Description

Model selection for multiple-trait composite multiple-interval mapping.

Usage

mtcmimAdd1(obj, y, x, xid, mpos, mdat, pp=1, len=1, type=1,
   iter=2500, tol=1e-8, ext=FALSE)

mtcmimDrop1(obj, y, x, xid, mpos, mdat, pp=1, len=1, type=1,
   iter=2500, tol=1e-8, ext=FALSE)

mtcmimStep(obj, y, x, xid, mpos, mdat, cv=0,
   direction=c("both", "backward", "forward"), pp=1, len=1,
   type=1, iter=2500, tol=1e-8, ext=FALSE)

Arguments

obj

object of class mtcmim.

y

n by p matrix, whose columns are dependent variables.

x

covariates; n by m numerical matrix.

xid

list of length p; xid[[j]] specifies columns of x as covariates for y[,j] .

mpos

data frame (id=marker index, ch=chromosome id, m=marker index on the chromosome, dist=genetic position in cM on the chromosome). Chromosome id should be an integer.

mdat

matrix of n rows; marker genotypes (1 or 0). columns should correspond to markers in the order.

pp

mapping population: BC-1, RIL-selfing-2, RIL-brother-sister-mating-3.

len

step length in search.

type

1 if traits can have the different sets of covariates and QTL, 2 if all have the same set of covariates and QTL.

ext

whether to perform an extensive search.

cv

critical value used in the likelihood ratio test to determine adding/dropping a QTL.

direction

forward selection, backward elimination or both directions.

iter

maximum number of iterations in a numerical process to estimate model parameters.

tol

convergence tolerance.

Value

• a list with the following components:
• loglik:
  log-likelihood of the final model
• a:
  covariate effects
• b:
  QTL effects
• sigma:
  residual variance-covariance
• qtl:
  QTL for each trait
• dists:
  QTL locations

See Also

mtcmim

Examples

data(etrait)
y<- traits[,1:5]
qtl<- vector("list",5); qtl[[1]]<- c(1,2)
eps<- data.frame(y=1,q1=1,q2=2)
dists<- dists[c(4,11),]
x<- mdat - 3/2
o<- mtcmim(y,mpos,mdat,dists=dists,
   qtl=qtl,eps=eps,win=5,range=-1,pp=2,len=1)
of<- mtcmimAdd1(o,y=y,mpos=mpos,mdat=mdat,pp=2,len=3)
os<- mtcmimStep(of,y=y,mpos=mpos,mdat=mdat,cv=25,
   direction="both",pp=2,len=3)