Create small squares on ggplot - r

Question

I'm trying to find a solution in order to create small squares figure that I will incorporate into a bigger figer. So for instance if we take this dataframe;
a b c d
1 1 1 1 1
2 1 0 1 1
I would like to create 2 plots :
for row1
and
for row2
Here is the data
data<-structure(list(a = c(1, 1),
b = c(1, 0), c = c(1,
1), d = c(1, 1
)), class = "data.frame", row.names = c(NA, -2L))
does someone have an idea please ?

Answer 1

The first step is to convert the data into a tidy (long) format:
library(tidyverse)
df1<-structure(list(a = c(1, 1),
b = c(1, 0), c = c(1,
1), d = c(1, 1
)), class = "data.frame", row.names = c(NA, -2L))
df1_tidy <- df1 %>%
rowid_to_column("rowID") %>%
pivot_longer(names_to = "colID", values_to= "value", -rowID) %>%
mutate(rowID = factor(rowID, levels = sort(unique(rowID), decreasing = T)))
Then to plot you can use geom_tile with coord_fixed to make sure you end up with squares.
ggplot(df1_tidy, aes(x = colID, y = rowID)) +
geom_tile(aes(width = value*0.75,height = value *0.75 , fill = colID)) +
coord_fixed() +
scale_fill_manual(values = c("#53ae32", "#2e76b5", "#f2d355", "#d23c28")) +
theme_void() +
theme(legend.position = "none")
Or in base R:
#genrate blank plot with correct dimensions
plot(1, type="n", xlab="", ylab="", xlim=c(0,5), ylim=c(-5,0), bty="n", xaxt='n',yaxt='n', ann=FALSE, )
clrs <- c("#53ae32", "#2e76b5", "#f2d355", "#d23c28")
for(i in 1:nrow(df1)){
for(j in 1:ncol(df1)){
cat(i,j, v, "\n")
v <- df1[i,j]
rect(xleft = j, xright = j+0.9, ybottom = -i, ytop = -i-0.9, col = ifelse(v==0, "white", clrs[j]), border = NA)
}
}

Answer 2

As to "is there a way without transforming to a long format" - well, not really for ggplot2. If you want to use ggplot, you will want to map your variables to an aesthetic, which will take columns, not rows. But there is no real issue here - you can just create a new data frame and add this to the other plot.
Another option instead of tiles is to draw points in the shape of squares. I take from your question that you want to indeed create separate plots. I am using patchwork here for the convenience of showing the output, but you can of course leave them separate. The advantage of this approach - you would not be constricted to use coord_fixed.
Apparently you also want to draw nothing when the value is 0. Best to replace with NA, so ggplot will ignore this point.
Thanks user GordonShumway for the colors!
library(ggplot2)
library(patchwork)
df<-structure(list(a = c(1, 1), b = c(1, 0), c = c(1, 1), d = c(1, 1 )), class = "data.frame", row.names = c(NA, -2L))
df_long <- data.frame(t(df))
df_long$colors <- c("#53ae32", "#2e76b5", "#f2d355", "#d23c28")
df_long[df_long == 0] <- NA
p_ls <-
purrr::map(1:nrow(df), function(i){
ggplot(df_long, aes_string(1:4, paste0("X", i))) +
geom_point(aes(color = colors), shape = 15, size = 20, show.legend = FALSE) +
scale_color_identity() +
coord_cartesian(clip = "off", expand = FALSE) +
theme_void() +
theme(plot.margin = margin(r = 1, l = 1, unit = "in")) +
ggtitle(paste("Plot", i))
})
wrap_plots(p_ls) +plot_layout(nrow = 2)
#> Warning: Removed 1 rows containing missing values (geom_point).
Created on 2021-03-12 by the reprex package (v1.0.0)