phensim: A function to simulated phenotypic data

Description

Simulates phenotypic data across arbitrary pedigrees. phensim simulate direct, maternal and paternal genetica and environmental effects for an arbitrary number of traits with arbitrary patterns of missing data.

Usage

phensim(pedigree, traits = 1, randomA = NULL, randomE = NULL, 
        parentalA = NULL, parentalE = NULL, sampled = NULL, 
              records = NULL, returnAllEffects = FALSE)

Arguments

pedigree

A pedigree

traits

The number of traits for which data should be simulated.

randomA

An additive genetic covariance matrix, with dimensions a multiple of traits - see details

randomE

An additive environmental covariance matrix, with dimensions a multiple of traits - see details

parentalA

A vector indicating which effects in randomA (if any) to treat as parental effects

parentalE

A vector indicating which effects in randomE (if any) to treat as parental effects

sampled

A vector indicating which individuals are sampled

records

A single value, array of matrix specifying data record availability - see details

returnAllEffects

If TRUE then all individual breeding values and environmental effects are returned

Value

phenotypes

A dataframe containing phenotypes for all indivduals specified to have records.

allEffects

(optional) A dataframe with all direct and indirect genetic and environmental effects.

Details

randomA and randomE are square matrices with dimension equal to the sum of the number direct and indirect effects. This must be a multiple of the number of traits, i.e. if an indirect effect is to be simulated for only one of multiple traits, those traits with no indirect effect should be included with (co)variances of zero.

parentalA and parentalE are optional vectors of characters indicating which trait positions in randomA and randomE are to be treated as indirect effects, and which effects to treat as maternal or paternal. Valid values are 'd', 'm', and 'p', for direct, maternal indirect and paternal indirect effects, respectively.

records can be specified either (1) as asingle value to be applied to all individuals and traits, (2) as a vector the same length as the number of traits, representing trait-specific rates to be applied uniformly to all individuals, or (3) as data frames with rows for each individual and columns for each trait. In the third option, observed patterns of data availability can be simulated by supplying 0s and 1s for missing and available individual genotypes, respectively.

References

Morrissey et al. 2007. Journal of Evolutionary Biology 20:2309-2321., Morrissey, M.B, and A.J. Wilson, 2009. pedantics, an R package for pedigree-based genetic simulation, and pedigree manipulation, characterisation, and viewing. Molecular Ecology Resources.

Examples

Run this code

# NOT RUN {
## make up a pedigree
id<-   c("a1","a2","a3","a4","a5","a6","a7","a8","a9")
dam<-  c(NA,NA,NA,"a1","a1","a1","a4","a4","a4")
sire<- c(NA,NA,NA,"a2","a2","a2","a5","a6","a6")
pedigree<-as.data.frame(cbind(id,sire,dam))

traits<-2
## no correlations
randomA<-diag(4)
randomE<-diag(4)
parentalA<-c("d","d","m","m")
parentalE<-c("d","d","m","m")

## generate phenoypic data based on this architecture
phensim(pedigree=pedigree,traits=2,randomA=randomA,randomE=randomE,
          parentalA=parentalA,parentalE=parentalE)

## let's do it again but see how the phenotypes were composed
phensim(pedigree=pedigree,traits=2,randomA=randomA,randomE=randomE,
          parentalA=parentalA,parentalE=parentalE,returnAllEffects=TRUE)
# }

Run the code above in your browser using DataCamp Workspace