besag.met {agridat} | R Documentation |
Multi-environment trial of corn laid out in incomplete-blocks at each location.
A data frame with 1152 observations on the following 7 variables.
county
county
row
row
col
column
rep
rep
block
incomplete block
yield
yield
gen
genotype, 1-64
Multi-environment trial of 64 corn hybrids in six counties in North Carolina. Each location had 3 replicates in in incomplete-block design with an 18x11 lattice of plots whose length-to-width ratio was about 2:1.
Note: In the original data, each county had 6 missing plots. This data has rows for each missing plot that uses the same county/block/rep to fill-out the row, sets the genotype to G01, and sets the yield to missing. These missing values were added to the data so that asreml could more easily do AR1xAR1 analysis using rectangular regions.
Each location/panel is:
Field length: 18 rows * 2 units = 36 units.
Field width: 11 plots * 1 unit = 11 units.
Julian Besag and D Higdon, 1999. Bayesian Analysis of Agricultural Field Experiments, Journal of the Royal Statistical Society: Series B (Statistical Methodology), 61, 691–746. Table 1.
http://doi.org/10.1111/1467-9868.00201
Retrieved from http://web.archive.org/web/19990505223413/www.stat.duke.edu/~higdon/trials/nc.dat
Used with permission of David Higdon.
data(besag.met) dat <- besag.met if(require(desplot)){ desplot(yield ~ col*row|county, dat, aspect=36/11, # true aspect out1=rep, out2=block, main="besag.met") } # Average reps datm <- aggregate(yield ~ county + gen, data=dat, FUN=mean) # Sections below fit heteroskedastic variance models (variance for each variety) # asreml takes 1 second, lme 73 seconds, SAS PROC MIXED 30 minutes # ---------------------------------------------------------------------------- ## Not run: # lme require(nlme) m1l <- lme(yield ~ -1 + gen, data=datm, random=~1|county, weights = varIdent(form=~ 1|gen)) m1l$sigma^2 * c(1, coef(m1l$modelStruct$varStruct, unc = FALSE))^2 ## G02 G03 G04 G05 G06 G07 G08 ## 91.90 210.75 63.03 112.05 28.39 237.36 72.72 42.97 ## ... etc ... ## End(Not run) # ---------------------------------------------------------------------------- ## Not run: # asreml3 require(asreml) # asreml Using 'rcov' ALWAYS requires sorting the data datm <- datm[order(datm$gen),] m1a <- asreml(yield ~ gen, data=datm, random = ~ county, rcov = ~ at(gen):units, predict=asreml:::predict.asreml(classify="gen")) require(lucid) vc(m1a)[1:7,] ## effect component std.error z.ratio constr ## county!county.var 1324 838.2 1.6 pos ## gen_G01!variance 91.93 58.82 1.6 pos ## gen_G02!variance 210.7 133.9 1.6 pos ## gen_G03!variance 63.03 40.53 1.6 pos ## gen_G04!variance 112.1 71.53 1.6 pos ## gen_G05!variance 28.39 18.63 1.5 pos ## gen_G06!variance 237.4 150.8 1.6 pos # We get the same results from asreml & lme plot(m1a$gammas[-1], m1l$sigma^2 * c(1, coef(m1l$modelStruct$varStruct, unc = FALSE))^2) # The following example shows how to construct a GxE biplot # from the FA2 model. dat <- besag.met dat <- transform(dat, xf=factor(col), yf=factor(row)) dat <- dat[order(dat$county, dat$xf, dat$yf), ] # First, AR1xAR1 m1 <- asreml(yield ~ county, data=dat, random = ~ gen:county, rcov = ~ at(county):ar1(xf):ar1(yf)) # Add FA1. # For ASExtras:::summary.fa, use fa(county,1):gen, NOT gen:fa(county,1) m2 <- update(m1, random=~fa(county,1):gen) # FA2 m3 <- update(m2, random=~fa(county,2):gen) # Use the loadings to make a biplot vars <- vc(m3) psi <- vars[grepl(".var$", vars$effect), "component"] la1 <- vars[grepl(".fa1$", vars$effect), "component"] la2 <- vars[grepl(".fa2$", vars$effect), "component"] mat <- as.matrix(data.frame(psi, la1, la2)) rot <- svd(mat[,-1])$v # rotation matrix lam <- mat[,-1] colnames(lam) <- c("load1", "load2") co3 <- coef(m3)$random # Scores are the GxE coefficients ix1 <- grepl("_Comp1:gen", rownames(co3)) ix2 <- grepl("_Comp2:gen", rownames(co3)) sco <- matrix(c(co3[ix1], co3[ix2]), ncol=2, byrow=FALSE) sco <- sco dimnames(sco) <- list(levels(dat$gen) , c('load1','load2')) rownames(lam) <- levels(dat$county) sco[,1] <- -1 * sco[,1] lam[,1] <- -1 * lam[,1] biplot(sco, lam, cex=.5, main="FA2 coefficient biplot") # G variance matrix gvar <- lam # Now get predictions and make an ordinary biplot p3 <- predict(m3, data=dat, classify="county:gen") p3 <- p3$pred$pval require("gge") bi3 <- gge(predicted.value ~ gen*county, data=p3, scale=FALSE) if(interactive()) dev.new() # Very similar to the coefficient biplot biplot(bi3, stand=FALSE, # what does 'stand' do? main="SVD biplot of FA2 predictions") # latent factor plots and more if(FALSE) { library(ASExtras) ASExtras:::summary.fa(m3, g.list=c("G01","G02","G03","G04","G05","G06","G07","G08")) out <- ASExtras:::summary.fa(m3,uniplot=0,blups=1,regplot=1,addedplot=0,heatmap=0) loads <- as.data.frame(out$gammas$`fa(county, 2):gen`$`rotated loads`) loads$county <- rownames(loads) regdat <- out$blups$`fa(county, 2):gen`$blups.inmet regdat <- merge(regdat, loads) library(latticeExtra) xyplot(blup ~ fac_1|gen, data=regdat, as.table=TRUE)+ xyplot(regblup ~ fac_1|gen, data=regdat, as.table=TRUE, type='r') } ## End(Not run) # ---------------------------------------------------------------------------- ## Not run: ## require(asreml4) ## # asreml Using 'rcov' ALWAYS requires sorting the data ## datm <- datm[order(datm$gen),] ## m1 <- asreml(yield ~ gen, data=datm, ## random = ~ county, ## resid = ~ dsum( ~ units|gen)) ## #summary(m1)$varcomp[1:7,] ## require(lucid) ## vc(m1)[1:7,] ## ## effect component std.error z.ratio bound ## ## county 1324 838.2 1.6 P 0 ## ## gen_G01(R) 91.95 58.85 1.6 P 0 ## ## gen_G02(R) 210.7 133.8 1.6 P 0.1 ## ## gen_G03(R) 63.04 40.55 1.6 P 0 ## ## gen_G04(R) 112.1 71.54 1.6 P 0 ## ## gen_G05(R) 28.38 18.61 1.5 P 0.1 ## ## gen_G06(R) 237.4 150.8 1.6 P 0 ## # We get the same results from asreml & lme ## plot(m1$vparameters[-1], ## m1l$sigma^2 * c(1, coef(m1l$modelStruct$varStruct, unc = FALSE))^2) ## # The following example shows how to construct a GxE biplot ## # from the FA2 model. ## dat <- besag.met ## dat <- transform(dat, xf=factor(col), yf=factor(row)) ## dat <- dat[order(dat$county, dat$xf, dat$yf), ] ## # First, AR1xAR1 ## m1 <- asreml(yield ~ county, data=dat, ## random = ~ gen:county, ## resid = ~ dsum( ~ ar1(xf):ar1(yf)|county)) ## # Add FA1 ## m2 <- update(m1, random=~gen:fa(county,1)) ## # FA2 ## m3 <- update(m2, random=~gen:fa(county,2)) ## # Use the loadings to make a biplot ## vars <- vc(m3) ## psi <- vars[grepl("!var$", vars$effect), "component"] ## la1 <- vars[grepl("!fa1$", vars$effect), "component"] ## la2 <- vars[grepl("!fa2$", vars$effect), "component"] ## mat <- as.matrix(data.frame(psi, la1, la2)) ## rot <- svd(mat[,-1])$v # rotation matrix ## lam <- mat[,-1] ## colnames(lam) <- c("load1", "load2") ## co3 <- coef(m3)$random # Scores are the GxE coefficients ## ix1 <- grepl("_Comp1$", rownames(co3)) ## ix2 <- grepl("_Comp2$", rownames(co3)) ## sco <- matrix(c(co3[ix1], co3[ix2]), ncol=2, byrow=FALSE) ## sco <- sco ## dimnames(sco) <- list(levels(dat$gen) , c('load1','load2')) ## rownames(lam) <- levels(dat$county) ## sco[,1] <- -1 * sco[,1] ## lam[,1] <- -1 * lam[,1] ## biplot(sco, lam, cex=.5, main="FA2 coefficient biplot") ## # G variance matrix ## gvar <- lam ## # Now get predictions and make an ordinary biplot ## p3 <- predict(m3, data=dat, classify="county:gen") ## p3 <- p3$pvals ## require("gge") ## bi3 <- gge(predicted.value ~ gen*county, data=p3, scale=FALSE) ## if(interactive()) dev.new() ## # Very similar to the coefficient biplot ## biplot(bi3, stand=FALSE, main="SVD biplot of FA2 predictions") ## End(Not run)