Adjusting figure margins using split.screen - r

Question

I am trying to produce multiple plots using the split.screen option and I need to have 7 plots on the page. One of them should be plotted on its own and the other 6 plotted repeatedly using a for loop.
This is my code for some simulation I am carrying out. It runs well, but I have two potential problems:
I am not sure which of the plots actually gets plotted because I couldn't get the assigned label to show up on the bigger plot.
The plot showing on screen 1 is not the actual data because I have plotted it separately and know what it should look like.
Simulating the data:
numpop = 2
N = 1250
nSNP = 5000
Fst = 0.001
omega = c(0.5, 0.5)
propnExtreme = 0.1
nsim = 10
Fst.obs = vector(length = nSNP)
pdiffs = vector(length = nSNP)
genomat = matrix(nrow = N, ncol = nSNP)
for (i in 1:nSNP){
p = runif(1, 0.1, 0.9)
alpha = p * (1 - Fst) / Fst
beta = (1 - p) * (1 - Fst) / Fst
ps = rbeta(numpop, shape1 = alpha, shape2 = beta)
vars = var(ps)
pdiffs[i] = max(ps) - min(ps)
Fst.obs[i] = vars / (p * (1 - p))
for (j in 1:numpop){
ind1 = (j-1) * N * omega[j] + 1
ind2 = j * N * omega[j]
freqs = c(ps[j]^2, 2 * ps[j] * (1 - ps[j]), (1 - ps[j])^2)
genomat[ind1:ind2, i] = sample(c(0, 1, 2), size = N*omega[j], replace = TRUE, prob = freqs)
}
}
snpmeans = colMeans(genomat)
pi = (1 + colSums(genomat)) / (2 + 2*nrow(genomat))
stdmat = scale(genomat, center=snpmeans, scale=sqrt(pi*(1-pi)))
pr = prcomp(stdmat, center=F, scale=F)
Plotting:
get( getOption("device" ) )()
png(file="myplot.png", width=2000, height = 1200)
par(oma = c(0,0,3,0))
split.screen(c(1,2)) # split display into two screens
plot(pr$x,
col = c(rep("red", N*omega[1]), rep("blue", N*omega[2])),
main = "Whole genotype data")
split.screen(c(2, 3), screen = 2) # now split the second into 2x3
for(i in 1:8) ## 8=#of screens
{
screen(i) # prepare screen i for output
fA=0.5
fa = 1-fA
combined_SNP <- sample(c(0:2), N, prob=c(fA^2, 2*fA*fa, fa^2), replace=T)
pheno_indep <-c()
##Phenotypes
for (i in 1:length(combined_SNP)){
if (combined_SNP[i] == '0') {
pheno_indep<- c(pheno_indep, rnorm(1, mean = 0.07, sd = 1))
} else if (combined_SNP[i ]== '1') {
pheno_indep <- c(pheno_indep, rnorm(1, mean = 0, sd = 1))
} else {
pheno_indep <- c(pheno_indep, rnorm(1, mean = -0.07, sd = 1))
}
}
l <- 1:N
combined_indep <- cbind(combined_SNP, pheno_indep, l)
sorted_combined <- combined_indep[order(combined_indep[, 2]), ]
##eps data
f = 0.1
Nums = nrow(sorted_combined)
keep <- c(1:(f*Nums), (Nums-(f*Nums)+1):Nums)
epsdat<- c(rep("0", f*Nums), rep("1", f*Nums))
EPS_dat <- as.factor(cbind(sorted_combined[keep, ], epsdat))
dim(EPS_dat) <- c(length(keep), 4)
#colnames(EPS_dat) <- c("Genotypes", "Phenotypes", "ID", "position")
PC_EPS <- prcomp((genomat[EPS_dat[, 3], ]))
plot(PC_EPS$x,
col=c(rep("red", f*Nums), rep("blue", f*Nums)))
}
close.screen(all=TRUE)
dev.off()
Result:
I have spent a lot of time trying to figure this out even with other packages like layout.show. Thanks!

Answer 1

Is the following what you expect to be plotted? (I added screen title to the small plots for illustration)
When you split the screens, you should have gotten the following on your console:
> split.screen(c(1, 2))
[1] 1 2
# (code used to plot first chart on the left)
> split.screen(c(2, 3), screen = 2)
[1] 3 4 5 6 7 8
As described in the help file ?split.screen, this is a a vector of screen numbers for the newly-created screens. So your valid screen numbers are 1 (already plotted), and 3-8 (6 small screens).
As such, the next line doesn't work as expected, since you're now looping through screens 1-8 rather than screens 3-8.
# instead of
for(i in 1:8) ## 8=#of screens
# use this
for(i in 3:8) ## 8=#of screens
As a side note, you should also use different loop counters for nested loops. Your outer loop (for the 6 small plots) used i as the loop counter. Within this loop, you have another loop for phenotypes, which used i as well. Since the screen selection was done at the start of each outer loop iteration, the code still worked in this case, but in general, best to keep the loop counters separate.