get_kernel: Envirotype-informed kernels for statistical models

In allogamous/EnvRtype: Envirotyping Tools for Crop Research and Genetic Improvment

Envirotype-informed kernels for statistical models

Description

Get multiple genomic and/or envirotype-informed kernels for bayesian genomic prediciton.

Usage

get_kernel(
  K_E = NULL,
  K_G = NULL,
  data = NULL,
  model = NULL,
  intercept.random = FALSE,
  reaction = FALSE,
  dimension_KE = NULL,
  ne = NULL,
  ng = NULL,
  env = "env",
  gid = "gid",
  y = "value"
)

Arguments

K_E	list. Contains n matrices of envirotype-related kernels (n x n genotypes-environment). If NULL, benchmarck genomic kernels are built.
K_G	list. Contains matrices of genomic enabled kernels (p x p genotypes). See BGGE::getK for more information.
data	data.frame. Should contain the following colunms: environemnt, genotype, phenotype.
model	character. Model structure for genomic predicion. It can be c('MM','MDs','EMM','EMDs'), in which MM (main effect model y = m + Z_{E} + G) and MDs (y = m + Z_{E} + G + GE).
intercept.random	boolean. Indicates the inclusion of a genomic random intercept (default = FALSE). For more details, see BGGE package vignette.
reaction	boolean. Indicates the inclusion of a reaction norm based GxE kernel (default = FALSE).
dimension_KE	character. size_E=c('q','n'). In the first, 'q' means taht the environmental relationship kernel has the dimensions of q x q observations,in which q is the number of environments. If 'n', the relationship kernel has the dimension n=pq, in which p is the number of genotypes
ne	numeric. denotes the number of environments (q)
ng	numeric. denotes the number of genotypes (p)
env	character. denotes the name of the column respectively to environments
gid	character. denotes the name of the column respectively to genotypes
y	character. denotes the name of the column respectively to phenotype values

Details

Detail on the model structures is given in the GitHub page: https://github.com/allogamous/EnvRtype/blob/master/Prediction.md

Value

A list of kernels (relationship matrices) to be used in genomic models.

Author(s)

Germano Costa Neto

See Also

env_typing W_matrix kernel_model

Examples

## Not run: 
### Loading the genomic, phenotype and weather data
data('maizeG'); data('maizeYield'); data("maizeWTH")

data("maizeYield") # toy set of phenotype data (grain yield per environment)
data("maizeG"    ) # toy set of genomic relationship for additive effects
data("maizeWTH")   # toy set of environmental data
y   = "value"      # name of the vector of phenotypes
gid = "gid"        # name of the vector of genotypes
env = "env"        # name of the vector of environments
ECs  = W_matrix(env.data = maizeWTH[maizeWTH$daysFromStart < 100, ], var.id = c("FRUE",'PETP',"SRAD","T2M_MAX"),statistic = 'mean')
## KG and KE might be a list of kernels
KE = list(W = env_kernel(env.data = ECs)[[2]])
KG = list(G=maizeG);
## Creating kernel models with get_kernel
## y = fixed + Genomic + error (main effect model, MM)
MM    = get_kernel(K_G = KG, y = y,gid = gid,env = env, data = maizeYield,model = "MM")
## y = fixed + Genomic + Genomic x Environment + error (MM plus a single GE deviation, MDs model, assuming GE as a block diagonal genomic effects)
MDs   = get_kernel(K_G = KG, y = y,gid = gid,env = env,  data = maizeYield, model = "MDs")
EMM   = get_kernel(K_G = KG, K_E = KE, y = y,gid = gid,env = env,  data = maizeYield, model = "EMM")
EMDs  = get_kernel(K_G = KG, K_E = KE, y = y,gid = gid,env = env,  data = maizeYield, model = "EMDs")
RMMM  = get_kernel(K_G = KG, K_E = KE, y = y,gid = gid,env = env,  data = maizeYield, model = "RNMM")
RNMDs = get_kernel(K_G = KG, K_E = KE, y = y,gid = gid,env = env,  data = maizeYield, model = "RNMDs")
## Examples of Models without any genetic relatedness, which G = I a identity for genotypes
MDs   = get_kernel(K_G = NULL, y = y,gid = gid,env = env,  data = maizeYield, model = "MDs")
EMDs  = get_kernel(K_G = NULL, K_E = KE, y = y,gid = gid,env = env,  data = maizeYield, model = "EMDs")

## End(Not run)

allogamous/EnvRtype documentation built on Nov. 1, 2024, 3:48 a.m.