Reorder Columns in graph made by ggplot

Question

I’m producing a bar chart, but the ggplot places the bars in alphabetical order. I need the bars to appear with the names in the order I define. How should I do?

Dice

nif <- structure(list(Genes = c("nifW", "nifV", "nifS", "nifU", "nifQ", 
"nifX", "nifN", "nifE", "nifK", "nifD", "nifH", "nifT", "nifZ", 
"nifZ2", "nifB", "nifA"), A = c(7.57, 6.15, 8.21, 9.36, 7.58, 
8.78, 7.49, 6.98, 8.86, 9.56, 8.24, 4.38, 4.96, 5.02, 8.27, 2.34
), B = c(2.81, 1.58, 4.17, NA, 3.46, 2.29, 4.39, 4, NA, 4.46, 
NA, 4.58, 3.81, 3.81, 3.32, 0.85), C = c(2.29, 2.74, 3, 3.71, 
3.58, 1, 1.93, 2.66, 5.78, 5.73, 7.17, 0.93, 2.29, 2.29, 4.17, 
-0.15), D = c(NA, 8.54, 7.07, NA, NA, 7.74, 7.31, 7.83, 7.75, 
8.26, 8.8, NA, NA, NA, 10.02, NA)), class = "data.frame", row.names = c(NA, 
-16L))

Code

library(dplyr)
library(ggplot2)
library(reshape2)

nif <- nif %>%
  mutate(Genes = factor(Genes, levels=c("nifW", "nifV", "nifS", "nifU", "nifQ","nifX","nifN","nifE", "nifK", "nifD", "nifH", "nifT", "nifZ", "nifZ2", "nifB", "nifA")))

df.long <- melt(nif)

ggplot(df.long,aes(x=Genes,y=value,fill=factor(variable))) +
     geom_bar(stat="identity", position="dodge") +
     scale_fill_discrete(name="Species") +
     ylab("Log2 Fold Change") + 
     theme(axis.text.x = element_text(angle = 90)) +
     theme(legend.text = element_text(colour="black", size = 10, face = "italic"))

Answer 1

I believe the code below can help you:

nif <- nif %>%
  mutate(Genes = factor(Genes, 
                   levels = c("nifW", "nifV", "nifS", "nifU", "nifQ","nifX","nifN","nifE", 
                              "nifK", "nifD", "nifH", "nifT", "nifZ", "nifZ2", "nifB", 
                              "nifA")))

df.long <- melt(nif, id.vars = "Genes")

ggplot(df.long, aes(x=Genes, y=value, fill=factor(variable))) +
  geom_bar(stat="identity", position="dodge") +
  scale_fill_discrete(name="Species") +
  ylab("Log2 Fold Change") + 
  theme(axis.text.x = element_text(angle = 90)) +
  theme(legend.text = element_text(colour="black", size = 10, face = "italic"))

Answer 2

One way is just to turn into factor after reformatting to long format. In the following code I will use the tidyr::pivot_longer and do everything in the same pipe.

library(dplyr)
library(tidyr)
library(ggplot2)

nif %>%
  pivot_longer(
    cols = -Genes,
    names_to = "variable",
    values_to = "value"
  ) %>% 
  mutate(Genes = factor(Genes, 
                        levels=c("nifW", "nifV", "nifS", "nifU", 
                                 "nifQ","nifX","nifN","nifE", "nifK", 
                                 "nifD", "nifH", "nifT", "nifZ", 
                                 "nifZ2", "nifB", "nifA"))) %>%
  ggplot(aes(x = Genes, y = value, fill = variable)) +
  geom_bar(stat = "identity", position = "dodge") +
  scale_fill_discrete(name = "Species") +
  ylab("Log2 Fold Change")) + 
  theme(axis.text.x = element_text(angle = 90),
        legend.text = element_text(colour = "black", size = 10, face = "italic"))

Note: Although I’m not in the question, if you want the "Log2" with the 2 in subscript, just change the ylab to the following.

ylab(expression(paste(Log[2] ~ "Fold Change")))