create.gpData: Create genomic prediction data object

In synbreed: Framework for the Analysis of Genomic Prediction Data using R

Description

This function combines all raw data sources in a single, unified data object of class gpData. This is a list with elements for phenotypic, genotypic, marker map, pedigree and further covariate data. All elements are optional.

Usage

create.gpData(
  pheno = NULL,
  geno = NULL,
  map = NULL,
  pedigree = NULL,
  family = NULL,
  covar = NULL,
  reorderMap = TRUE,
  map.unit = "cM",
  repeated = NULL,
  modCovar = NULL,
  na.string = "NA",
  cores = 1
)

Arguments

pheno	data.frame with individuals organized in rows and traits organized in columns. For unrepeated measures unique rownames should identify individuals. For repeated measures, the first column identifies individuals and a second column indicates repetitions (see also argument repeated).
geno	matrix with individuals organized in rows and markers organized in columns. Genotypes could be coded arbitrarily. Missing values should be coded as NA. Colums or rows with only missing values not allowed. Unique rownames identify individuals and unique colnames markers. If no rownames are available, they are taken from element pheno (if available and if dimension matches). If no colnames are used, the rownames of map are used if dimension matches.
map	data.frame with one row for each marker and two columns (named chr and pos). First columns gives the chromosome (numeric or character but not factor) and second column the position on the chromosome in centimorgan or the physical distance relative to the reference sequence in basepairs. Unique rownames indicate the marker names which should match with marker names in geno. Note that order and number of markers must not be identical with the order in geno. If this is the case, gaps in the map are filled with NA to ensure the same number and order as in element geno of the resulting gpData object.
pedigree	Object of class pedigree.
family	data.frame assigning individuals to families with names of individuals in rownames This information could be used for replacing of missing values with function codeGeno.
covar	data.frame with further covariates for all individuals that either appear in pheno, geno or pedigree$ID, e.g. sex or age. rownames must be specified to identify individuals. Typically this element is not specified by the user.
reorderMap	logical. Should markers in geno and map be reordered by chromosome number and position within chromosome according to map (default = TRUE)?
map.unit	character. Unit of position in map, i.e. 'cM' for genetic distance or 'bp' for physical distance (default = 'cM').
repeated	This column is used to identify the replications of the phenotypic values. The unique values become the names of the third dimension of the pheno object in the gpData. This argument is only required for repeated measurements.
modCovar	vector with colnames which identify columns with covariables in pheno. This argument is only required for repeated measurements.
na.string	character or vector of characters. You can specify values with which NA is coded in your geno object. In case you read missing values from a file not as missing, but as character strings. It can be specified more than one value for missings in a vector. Default is "NA".
cores	numeric. Here you can specify the number of cores you like to use.

Details

The class gpData is designed to provide a unified framework for data related to genomic prediction analysis. Every data source can be omitted. In this case, the corresponding argument must be NULL. By default (argument reorderMap), markers in geno are ordered by their position in map. Individuals are ordered in alphabetical order.

An object of class gpData can contain different subsets of individuals or markers in the elements pheno, geno and pedigree. In this case the id in covar comprises all individuals that either appear in pheno, geno and pedigree. Two additional columns in covar named phenotyped and genotyped are automatically generated to identify individuals that appear in the corresponding gpData object.

Value

Object of class gpData which is a list with the following elements

covar	data.frame with information on individuals
pheno	array (individuals x traits x replications) with phenotypic data
geno	matrix marker matrix containing genotypic data. Columns (marker) are in the same order as in map (if reorderMap=TRUE.)
pedigree	object of class pedigree
map	data.frame with columns 'chr' and 'pos' and markers sorted by 'pos' within 'chr'
phenoCovars	array with phenotypic covariates
info	list with additional information on data (coding of data, unit in map) From synbreed version 0.11-11 on the function codeGeno adds here the package version which was used to do the coding. There are differences in codings between version 0.10-11 and 0.11-0!

Note

In case of missing row names or column names in one item, information is substituted from other elements (assuming the same order of individuals/markers) and a warning specifying the assumptions is returned. Please check them carefully.

Author(s)

Valentin Wimmer and Hans-Juergen Auinger with contributions be Peter VandeHaar

See Also

codeGeno, summary.gpData, gpData2data.frame

Examples

set.seed(123)
# 9 plants with 2 traits
n <- 9 # only for n > 6
pheno <- data.frame(Yield = rnorm(n, 200, 5), Height = rnorm(n, 100, 1))
rownames(pheno) <- letters[1:n]

# marker matrix
geno <- matrix(sample(c("AA", "AB", "BB", NA),
  size = n * 12, replace = TRUE,
  prob = c(0.6, 0.2, 0.1, 0.1)
), nrow = n)
rownames(geno) <- letters[n:1]
colnames(geno) <- paste("M", 1:12, sep = "")

# genetic map
# one SNP is not mapped (M5) and will therefore be removed
map <- data.frame(chr = rep(1:3, each = 4), pos = rep(1:12))
map <- map[-5, ]
rownames(map) <- paste("M", c(1:4, 6:12), sep = "")

# simulate pedigree
ped <- simul.pedigree(3, c(3, 3, n - 6))

# combine in one object
gp <- create.gpData(pheno, geno, map, ped)
summary(gp)


# 9 plants with 2 traits , 3 replications
n <- 9 #
pheno <- data.frame(
  ID = rep(letters[1:n], 3), rep = rep(1:3, each = n),
  Yield = rnorm(3 * n, 200, 5), Height = rnorm(3 * n, 100, 1)
)

# combine in one object
gp2 <- create.gpData(pheno, geno, map, repeated = "rep")
summary(gp2)

synbreed documentation built on March 12, 2021, 3:01 a.m.