Boxplot or ggplot

Question

[Kai_Qi]

I am comparing gene expression level across 4 developmental stages: E11, E14, E18, Adult. I have about 800 genes selected. Each genes has a read at these satages. Now I made a matrix: the row name are gene ID, while the column name are the 4 stages. each columns contains the read number. I can easily use Boxplot function to get the figure, but I found that if I want to do t-test to see the p value and add them to the boxplot a lot of online solutions are ggplot.

I thought about ggplot, the problem is that if I use ggplot in the beginning I have to put a new matirx with 4X800 rows? it looks pretty intimidating.

Is my understanding right? Any advice is appreciated.

Answer 1

[cpad0112]

I thought about ggplot, the problem is that if I use ggplot in the beginning I have to put a new matirx with 4X800 rows? it looks pretty intimidating.

It is not intimidating once you understand the basics of format conversion. Currently your data is in wide format (800 genes x 4 stages). Data needs to be rearranged in a different format for use in ggplot and the format is called long format (3 columns - Genes, Stages and Counts). The reformat is done by single line code using gather function in tidyr library in R.

Let us say if your data is in following format:

> df
     genes E11  E14  E18 Adult
1   gene_1 526  555  772   818
2   gene_2 286  555 1077  1099
3   gene_3 468  433  937   951
4   gene_4 584  618 1001   970
5   gene_5 292  700  821   942
6   gene_6 334  526  803   901
7   gene_7 214  683  922   914
8   gene_8 558 1102  738   799
9   gene_9 550  494  799   920
10 gene_10 581  993  991   996

You need to run following code to create a data frame for working with ggplot:

library(tidyr)
df2=gather(df,"Stages","Reads",-genes)

Output would be:

> as_tibble(df2)
# A tibble: 40 x 3
   genes   Stages Reads
   <chr>   <chr>  <int>
 1 gene_1  E11      526
 2 gene_2  E11      286
 3 gene_3  E11      468
 4 gene_4  E11      584
 5 gene_5  E11      292
 6 gene_6  E11      334
 7 gene_7  E11      214
 8 gene_8  E11      558
 9 gene_9  E11      550
10 gene_10 E11      581
# … with 30 more rows

Use df2 for ggplot functions.

Answer 2

[MatthewP]

Yes you are right, your new column name should be Read and Stage. You can use function gather to generate new matrix. Below is a simple example.

> library(tidyverse, quietly = TRUE)
> matrix1 <- tibble(a = c(0, 0, 0), b = 1:3, c = 1:3)
> matrix1
# A tibble: 3 x 3
      a     b     c
  <dbl> <int> <int>
1     0     1     1
2     0     2     2
3     0     3     3


 > tidyr::gather(matrix1, key = "group", value = "new_col", a, b, c)
# A tibble: 9 x 2
  group new_col
  <chr>   <dbl>
1 a           0
2 a           0
3 a           0
4 b           1
5 b           2
6 b           3
7 c           1
8 c           2
9 c           3

[Kai_Qi]

mydat2 <- tidyr::gather(mydat, key = "stage", value = "new_col", E11, E14, E18, Adult)

mydat is a matrix, I chekced all the numbers in E11, E14, E18, Adult are numeric. but got an error:

Error in gather_(data, key_col = compat_as_lazy(enquo(key)), value_col = compat_as_lazy(enquo(value)),  : 
  unused arguments (E14, E18, Adult)