How to plot a table with multiple columns as a box plot

I am trying to plot a box plot with the Trinucleotide as the x axis (so 64 trinucleotides on the x axis) and the frequency of each trinucleotide in each of 6 samples then color code the plot according to the sample. This is a snippet of the table and the code I have so far as well as the type of graph I want.

ibrary(tidyverse)
library(readxl)

marte - read_xlsx(TrinucleotideFrequency06182021.xlsx)
marte - gather (marte, xzl.mmu.C57.testis.wt.adult.40S_crosslink.rep1+rept1.RPF.trimmed.gz.x_rRNA.x_hairpin.mm10v1.unique.+jxn.bed13.40S.sense.hybrid.utr3.1up.5end.PNLDC1.rep1.bed6, xzl.mmu.C57.testis.wt.adult.40S_crosslink.rep2+rept2.RPF.R1.trimmed.gz.x_rRNA.x_hairpin.mm10v1.unique.+jxn.bed13.40S.sense.hybrid.utr3.1up.5end.PNLDC1.rep1.bed6, xzl.mmu.C57.testis.wt.adult.40S_crosslink.rep3+rept3.RPF.R1.trimmed.gz.x_rRNA.x_hairpin.mm10v1.unique.+jxn.bed13.40S.sense.hybrid.utr3.1up.5end.PNLDC1.rep1.bed6, xzl.mmu.C57.testis.wt.adult.80S_crosslink.rep1+rept1.RPF.trimmed.gz.x_rRNA.x_hairpin.mm10v1.unique.+jxn.bed13.RPF.sense.hybrid.utr3.1up.5end.PNLDC1.rep1.bed6, xzl.mmu.C57.testis.wt.adult.80S_crosslink.rep2+rept2.RPF.R1.trimmed.gz.x_rRNA.x_hairpin.mm10v1.unique.+jxn.bed13.RPF.sense.hybrid.utr3.1up.5end.PNLDC1.rep1.bed6, xzl.mmu.C57.testis.wt.adult.80S_crosslink.rep3+rept3.RPF.R1.trimmed.gz.x_rRNA.x_hairpin.mm10v1.unique.+jxn.bed13.RPF.sense.hybrid.utr3.1up.5end.PNLDC1.rep1.bed6,key=gene, value=value)
marte$gene - as.factor(marte$Trinucleotide)  
marte$group - as.factor(marte$gene)

ggplot(marte, aes(x = gene, y = value, color = group)) +
  geom_boxplot()

head (marte) produces the output below and underneath is a part of the table I am using to generate the plot with the first column as the trinucleotide and the other 6 as each gene respectively.

 Trinucleotide gene   value group
  chr         fct  dbl fct
1 AAA           AAA   0.0214 AAA  
2 TAG           TAG   0.0199 TAG  
3 AGC           AGC   0.0132 AGC  
4 TGT           TGT   0.0338 TGT  
5 GAT           GAT   0.0130 GAT  
6 CAC           CAC   0.0112 CAC 

AAA 0.021383202 0.016654469 0.022484448 0.025311535 0.025495724 0.02017925
TAG 0.019927531 0.018790672 0.015649845 0.02230479  0.019363723 0.02017925
AGC 0.013209711 0.016891825 0.014520044 0.014302046 0.014804269 0.012904701
TGT 0.033757832 0.034990209 0.031899462 0.034446096 0.034475058 0.031047513
GAT 0.013006661 0.007615167 0.009163947 0.010540757 0.008234933 0.012000828
CAC 0.011210666 0.015012758 0.01241387  0.011478221 0.011045046 0.013579884
GGG 0.012712095 0.011927132 0.013222864 0.011558249 0.014292494 0.014647108
CGA 0.00230219  0.000969203 0.001325076 0.002034983 0.001656292 0.002036438
TCG 0.004169681 0.007377811 0.005649008 0.005133189 0.006057561 0.006131094
ACA 0.011259284 0.013825979 0.010656401 0.009900537 0.010617015 0.009670358

Topic ggplot2 rstudio r

Category Data Science

1
answered at

Based on the explanations in the comments, what you need is to format the data so that every sample has the right 'group' label. Since there are only 6 columns I would just do it manually like below.

marte <- read_xlsx("TrinucleotideFrequency06182021.xlsx")

a40 <- data.frame(gene=marte$Trinucleotide,value=marte[,2])
b40 <- data.frame(gene=marte$Trinucleotide,value=marte[,3])
c40 <- data.frame(gene=marte$Trinucleotide,value=marte[,4])
a80 <- data.frame(gene=marte$Trinucleotide,value=marte[,5])
b80 <- data.frame(gene=marte$Trinucleotide,value=marte[,6])
c80 <- data.frame(gene=marte$Trinucleotide,value=marte[,7])

# these should not be necessary:
colnames(a40) <- c('gene','value')
colnames(b40) <- c('gene','value')
colnames(c40) <- c('gene','value')
colnames(a80) <- c('gene','value')
colnames(b80) <- c('gene','value')
colnames(c80) <- c('gene','value')

gp1 <- rbind(a40,b40,c40)
gp1$group <- rep('40s',nrow(gp1))
gp2 <- rbind(a80,b80,c80)
gp2$group <- rep('80s',nrow(gp2))

ggplot(marte, aes(x = gene, y = value, color = group)) +geom_boxplot()+ theme(axis.text.x = element_text(angle = 90))+coord_flip()

Plot

Explanation: for each sample a dataframe which contains only the values coming from the three columns of this sample. The three parts are concatenated by rows (rbind) with the columns gene and value, then the column group is added containing the id of the group (sample) in every row.

For the record, doing boxplots with only two values is not great, a boxplot is supposed to show the distribution of a set of values. I would suggest simple barplots instead, with the two values shown as different bars.

About

Geeks Mental is a community that publishes articles and tutorials about Web, Android, Data Science, new techniques and Linux security.