| Type: | Package |
| Title: | Hierarchical Spatial Finlay-Wilkinson Model |
| Version: | 0.1.0 |
| Author: | Xingche Guo <xguo@iastate.edu> |
| Maintainer: | Xingche Guo <xguo@iastate.edu> |
| Description: | Estimation and Prediction Functions Using Bayesian Hierarchical Spatial Finlay-Wilkinson Model for Analysis of Multi-Environment Field Trials. |
| License: | GPL-3 |
| Encoding: | UTF-8 |
| LazyData: | true |
| RoxygenNote: | 6.1.1 |
| Depends: | R (≥ 2.10) |
| LinkingTo: | Rcpp, RcppArmadillo |
| Imports: | Rcpp |
| NeedsCompilation: | yes |
| Packaged: | 2022-03-29 07:41:56 UTC; apple |
| Repository: | CRAN |
| Date/Publication: | 2022-03-29 18:40:02 UTC |
Estimation Function for Hierarchical Finlay-Wilkinson Model
Description
This function ignores spatial effects.
Usage
HFWM_est(Y, VAR, ENV, kin_info = FALSE, A = NULL, env_info = FALSE,
Z = NULL, inits = NULL, hyper_para = NULL, M_iter = 5000,
burn_in = 3000, thin = 5, save_chain = FALSE, seed = NULL)
Arguments
Y |
A length-N numerical response vector |
VAR |
A length-N factor/character vector indicating the genotype information of Y |
ENV |
A length-N factor/character vector indicating the field information of Y |
kin_info |
A logical parameter controling if to use kinship matrix |
A |
kinship matrix, give value only if kin_info = TRUE |
env_info |
A logical parameter controling whether to use environmental covariates |
Z |
environmental covariates matrix with rownames = field names, give value only if env_info = TRUE |
inits |
initial values, default is given |
hyper_para |
hyper-parameter values, default is given |
M_iter |
Total iteration number |
burn_in |
Burn in number |
thin |
Thinning value |
save_chain |
A logical parameter controling whether to save MCMC chain: 'Chains.rds' in current working directory |
seed |
Random seed value |
Value
Mean estimates and RMSE value
Examples
library(spFW)
# load data
data(spFW_example_data)
Y <- spFW_example_data$yield
VAR <- spFW_example_data$geno
ENV <- spFW_example_data$loc
COOR <- spFW_example_data[,c(4,5)]
# run model
fit0 <- HFWM_est(Y, VAR, ENV, M_iter = 1000, burn_in = 500, thin = 5)
# plot estimated Y
plot(Y, fit0$yhat)
Prediction Function for Hierarchical Finlay-Wilkinson Model
Description
This function ignores spatial effects.
Usage
HFWM_pred(Y, VAR, ENV, VAR2, ENV2, save_int = FALSE, kin_info = FALSE,
A = NULL, inits = NULL, hyper_para = NULL, M_iter = 5000,
burn_in = 3000, thin = 5, seed = NULL)
Arguments
Y |
A length-N1 numerical response vector from training set |
VAR |
A length-N1 factor/character vector indicating the genotype information of Y |
ENV |
A length-N1 factor/character vector indicating the field information of Y |
VAR2 |
A length-N2 factor/character vector indicating the genotype information of testing set |
ENV2 |
A length-N2 factor/character vector indicating the field information of of testing set |
save_int |
A logical parameter controling whether to save prediction credible intervals |
kin_info |
A logical parameter controling if to use kinship matrix |
A |
kinship matrix, give value only if kin_info = TRUE |
inits |
initial values, default is given |
hyper_para |
hyper-parameter values, default is given |
M_iter |
Total iteration number |
burn_in |
Burn in number |
thin |
Thinning value |
seed |
Random seed value |
Value
Mean prediction values and/or prediction intervals
Examples
library(spFW)
# load and split data
data(spFW_example_data)
idx_pred <- sample(125, 25)
Y0 <- spFW_example_data$yield
VAR0 <- spFW_example_data$geno
ENV0 <- spFW_example_data$loc
Y1 <- Y0[-idx_pred]
Y2 <- Y0[idx_pred]
VAR1 <- VAR0[-idx_pred]
VAR2 <- VAR0[idx_pred]
ENV1 <- ENV0[-idx_pred]
ENV2 <- ENV0[idx_pred]
order_y <- order(Y2)
# run model
pred0 <- HFWM_pred(Y1, VAR1, ENV1, VAR2, ENV2, save_int = TRUE,
M_iter = 1000, burn_in = 500, thin = 5)
# visualize prediction results
plot(1:25, pred0$PY[order_y], ylim = c(50, 250), pch = 15, col = "red",
xlab = "Plant ID for Prediction", ylab = "Yield",
main = "95% Prediction Intervals with Predicted Mean (Red) Versus True Yield (Blue)")
points(1:25, Y2[order_y], col = "blue")
for (i in 1:25){
lines(x = c(i,i), y = c(pred0$PY_CI[,order_y][1,i], pred0$PY_CI[,order_y][4,i]))
}
Estimation Function for Hierarchical Spatial Finlay-Wilkinson Model
Description
This function considers spatial adjustments.
Usage
HSFWM_est(Y, VAR, ENV, COOR, kin_info = FALSE, A = NULL,
env_info = FALSE, Z = NULL, inits = NULL, hyper_para = NULL,
M_iter = 5000, burn_in = 3000, thin = 5, save_chain = FALSE,
seed = NULL)
Arguments
Y |
A length-N numerical response vector |
VAR |
A length-N factor/character vector indicating the genotype information of Y |
ENV |
A length-N factor/character vector indicating the field information of Y |
COOR |
A N by 2 numerical matrix indicating the spatial locations of Y |
kin_info |
A logical parameter controling if to use kinship matrix |
A |
kinship matrix, give value only if kin_info = TRUE |
env_info |
A logical parameter controling whether to use environmental covariates |
Z |
environmental covariates matrix with rownames = field names, give value only if env_info = TRUE |
inits |
initial values, default is given |
hyper_para |
hyper-parameter values, default is given |
M_iter |
Total iteration number |
burn_in |
Burn in number |
thin |
Thinning value |
save_chain |
A logical parameter controling whether to save MCMC chain: 'Chains.rds' in current working directory |
seed |
Random seed value |
Value
Mean estimates and RMSE value
Examples
library(spFW)
# load data
data(spFW_example_data)
Y <- spFW_example_data$yield
VAR <- spFW_example_data$geno
ENV <- spFW_example_data$loc
COOR <- spFW_example_data[,c(4,5)]
# run model
fit1 <- HSFWM_est(Y, VAR, ENV, COOR,
M_iter = 1000, burn_in = 500, thin = 5)
# plot estimated Y
plot(Y, fit1$yhat)
Prediction Function for Hierarchical Spatial Finlay-Wilkinson Model
Description
This function considers spatial adjustments.
Usage
HSFWM_pred(Y, VAR, ENV, COOR, VAR2, ENV2, COOR2, save_int = FALSE,
kin_info = FALSE, A = NULL, inits = NULL, hyper_para = NULL,
M_iter = 5000, burn_in = 3000, thin = 5, seed = NULL)
Arguments
Y |
A length-N1 numerical response vector from training set |
VAR |
A length-N1 factor/character vector indicating the genotype information of Y |
ENV |
A length-N1 factor/character vector indicating the field information of Y |
COOR |
A N1 by 2 numerical matrix indicating the spatial locations of Y |
VAR2 |
A length-N2 factor/character vector indicating the genotype information of testing set |
ENV2 |
A length-N2 factor/character vector indicating the field information of of testing set |
COOR2 |
A N2 by 2 numerical matrix indicating the spatial locations of testing set |
save_int |
A logical parameter controling whether to save prediction credible intervals |
kin_info |
A logical parameter controling if to use kinship matrix |
A |
kinship matrix, give value only if kin_info = TRUE |
inits |
initial values, default is given |
hyper_para |
hyper-parameter values, default is given |
M_iter |
Total iteration number |
burn_in |
Burn in number |
thin |
Thinning value |
seed |
Random seed value |
Value
Mean prediction values and/or prediction intervals
Examples
library(spFW)
# load and split data
data(spFW_example_data)
idx_pred <- sample(125, 25)
Y0 <- spFW_example_data$yield
VAR0 <- spFW_example_data$geno
ENV0 <- spFW_example_data$loc
COOR0 <- spFW_example_data[,c(4,5)]
Y1 <- Y0[-idx_pred]
Y2 <- Y0[idx_pred]
VAR1 <- VAR0[-idx_pred]
VAR2 <- VAR0[idx_pred]
ENV1 <- ENV0[-idx_pred]
ENV2 <- ENV0[idx_pred]
COOR1 <- COOR0[-idx_pred,]
COOR2 <- COOR0[idx_pred,]
order_y <- order(Y2)
# run model
pred1 <- HSFWM_pred(Y1, VAR1, ENV1, COOR1, VAR2, ENV2, COOR2, save_int = TRUE,
M_iter = 1000, burn_in = 500, thin = 5)
# visualize prediction results
plot(1:25, pred1$PY[order_y], ylim = c(50, 250), pch = 15, col = "red",
xlab = "Plant ID for Prediction", ylab = "Yield",
main = "95% Prediction Intervals with Predicted Mean (Red) Versus True Yield (Blue)")
points(1:25, Y2[order_y], col = "blue")
for (i in 1:25){
lines(x = c(i,i), y = c(pred1$PY_CI[,order_y][1,i], pred1$PY_CI[,order_y][4,i]))
}
Example Data for Analysis
Description
A data frame containing the plant yield, genotypes, environments, and within-field positions.
Usage
spFW_example_data
Format
A data frame with 5 elements, which are:
- yield
Plant yield
- geno
Plant genotype ID
- loc
Plant environment ID
- range
Plant x-coordinate at a given environment
- pass
Plant y-coordinate at a given environment