Btukfyl

I am using R studio to plot boxplot.

My data looks like --

level  Q9HC84       P02768

2      232742108.4  32822363539

2      2886647577   25661800126

1      629319882.2  26745385141

1      1293306218   2173862517

0      632315638.5  994599604.8

0      221394201.3  1659282642

Here is the command I am using to plot a graph:

boxplot(Q9HC84 ~ level ,  data = x , xlab = "Cases", ylab = "Protein",

        main = "Q9HC84" , notch = FALSE, 

        varwidth = TRUE, names = c("C","T","O"))

This shows me a perfect result that is what I want.
But I have 90 proteins name in the data and I want to change this script in a manner that it will generate boxplot for each protein separately.

Same thing i did with lineplot that was easy, but in boxplot it's hard to find how to apply loop. I tried this script ----

for(i in 3:5)

{

  x = data.frame(proteomics$level)

  y = data.frame(proteomics[i])

boxplot(y ~ x ,  xlab = "Cases", ylab = "Metabolite",

        main = "Q9HC84" , notch = FALSE, 

        varwidth = TRUE, names = c("Control","Tobacco","OSCC"))



}

can anybody help me ?
Thank you in advance

It's difficult to create boxplots with this many groups while still having the plot being readable.

# generate example data

set.seed(1)

np <- 20

level <- rep(1:3, times=c(4, 5, 5))

pnames <- replicate(

  np, paste(c(sample(LETTERS, 1), sample(c(0:9, LETTERS), 5)), collapse="")

)

prot <- round(runif(np*length(level), 2213942000, 328223635400), 1)

x <- as.data.frame(cbind(level, matrix(prot, ncol=np)))

colnames(x)[-1] <- pnames





# reshape to long format

library(reshape2)

x.long <- melt(x, id.vars="level")

x.long$level <- c("C","T","O")[x.long$level]



# plot vertically, level + variable

par(xaxs="i", yaxs="i", mar=c(3.5, 3, 0.5, 0.5))

col <- c("orange", "cornflowerblue", "purple")

boxplot(value ~ level + variable, data=x.long, cex=0.5, cex.axis=0.5, las=2, 

  col=col, border=col, boxwex=0.5)



# plot horizontally, variable + level

par(xaxs="i", yaxs="i", mar=c(2, 3.5, 0.5, 0.5))

col <- rep(c("orange", "cornflowerblue", "purple"), 

  each=length(unique(x.long$variable)))

boxplot(value ~ variable + level, data=x.long, cex=0.5, cex.axis=0.5, las=1,

  horizontal=TRUE, col=col, border=col, boxwex=0.5)

Here are 2 solutions:

Assuming your data (slightly simplified) look like this (with three proteins, but that number can obviously be extended):

df <- data.frame(

  level = c(2,2,1,1,0,0),

  prot1 = c(rnorm(6)+1),

  prot2 = c(rnorm(6)),

  prot3 = c(rnorm(6)-1)

)

you can re-structure the data in such a way that the protein names form one column and the values they have form another column:

df_new <- data.frame(

  level = rep(c(2,2,1,1,0,0),3),

  protein_names = c(rep("prot1",6),rep("prot2",6),rep("prot3",6)),

  protein_values = c(df$prot1, df$prot2, df$prot3)

)

df_new

The first solution is an R base solution:
This line of code plots all boxplots within a single frame:

boxplot(df_new$protein_values ~ df_new$level+df_new$protein_names)

This for loop plots a separate frame for each protein:

par(mfrow=c(1,3))

for(i in unique(df_new$protein_names)){

  boxplot(df_new[df_new$protein_names==i,]$protein_values ~   df_new[df_new$protein_names==i,]$level)

}

The second solution is a ggplot2 solution:

library(ggplot2)



ggplot(df_new, aes(as.factor(level), protein_values)) +

geom_boxplot() +

facet_grid(.~protein_names)