MasterBayes (version 2.58)

beta.loglik: Log-Likelihood of Beta

Description

Log-likelihood of beta given a pedigree and phenotypic data. Beta is the parameter vector for the multinomial log-linear model. Intended to be used within the function MLE.beta

Usage

beta.loglik(X, dam_pos=NULL, sire_pos=NULL, par_pos=NULL, beta=NULL,
   beta_map=NULL, merge=NULL, mergeN=NULL, nUS=c(0,0), shrink=NULL)

Arguments

X

list of design matrices for each offspring. Each element should either have dam (D) and/or sire (S) matrices, or a composite Dam/Sire (DS) matrix. See varPed for model types

dam_pos

position of each offspring's mother in the dam design matrix

sire_pos

position of each offspring's mother in the sire design matrix

par_pos

position of each offspring's parents in the composite dam/sire matrix

beta

parameter vector

beta_map

vector that maps beta onto the design matrices (see getXlist)

merge

optional vector that indicates columns of for which the parameter is transformed using the argument merge in varPed

mergeN

optional list of matrices for each offspring the columns of which refer to merged variables and the rows to the number of individuals that fall into each category defined by merge)

nUS

vector of the number of unsampled females and males, respectively. Only required if unsampled individuals have known phenotype.

shrink

optional scalar for the variance defining the ridge-regression likelihood penalisation.

Value

log-likelihood of beta given the pedigree and X.

References

Hadfield J.D. et al (2006) Molecular Ecology 15 3715-31 Smouse P.E. et al (1999) Journal of Evolutionary Biology 12 1069-1077

See Also

MLE.beta, MCMCped, varPed, getXlist

Examples

# NOT RUN {
data(WarblerP)
data(WarblerG)

GdP<-GdataPed(WarblerG)

res1<-expression(varPed("offspring", relational=FALSE, restrict=0))
var1<-expression(varPed(c("lat", "long"), gender="Male", 
  relational="OFFSPRING"))
res2<-expression(varPed("terr", gender="Female", relational="OFFSPRING", 
  restrict="=="))

PdP<-PdataPed(formula=list(var1,res1,res2), data=WarblerP)

# probability of paternity is modelled as a function of  distance 

X.list<-getXlist(PdP=PdP, GdP=GdP)

ped<-MLE.ped(X.list)$P

# get ML pedigree from genetic data alone

X<-lapply(X.list$X, function(x){list(S=x$XSs)})

# Extract Design matrices for Sires 

sire_pos<-match(ped[,3][as.numeric(names(X))], X.list$id)
sire_pos<-mapply(function(x,y){match(x, y$sire.id)}, sire_pos, X.list$X)

# row number of each design matrix corresponding to the ML sire. 

beta<-seq(-0.065,-0.0325, length=100)
beta_Loglik<-1:100
  for(i in 1:100){
     beta_Loglik[i]<-beta.loglik(X, sire_pos=sire_pos, beta=beta[i], 
     beta_map=X.list$beta_map) 
  }

plot(beta_Loglik~beta, type="l", main="Profile Log-likelihood for beta")
# }