Getting started with plater

Sean Hughes

2021-01-04

How plater helps you

plater makes it easy to work with data from experiments performed in plates.

Many scientific instruments (such as plate readers and qPCR machines) produce data in tabular form that mimics a microtiter plate: each cell corresponds to a well as physically laid out on the plate. For experiments like this, it’s often easiest to keep records of what was what (control vs. treatment, concentration, sample type, etc.) in a similar plate layout form. But while plate-shaped data is easy to think about, it’s not easy to analyze. The point of plater is to seamlessly convert plate-shaped data (easy to think about) into tidy data (easy to analyze). It does this by defining a simple, systematic format for storing information in plate layouts. Then it painlessly rearranges data that intuitive format into a tidy data frame.

There are just two steps:

  1. Put the data in a file in plater format
  2. Read in the data plater functions

The example

Imagine you’ve invented two new antibiotics. To show how well they work, you filled up a 96-well plate with dilutions of the antibiotics and mixed in four different types of bacteria. Then, you measured how many of the bacteria got killed. So for each well in the plate you know:

The first three items are variables you chose in setting up the experiment. The fourth item is what you measured.

Step 1: Put the data in plater format

The first step is to create a file for the experiment. plater format is designed to store all the information about an experiment in one file. It’s simply a .csv file representing a single plate, containing one or more plate layouts. Each layout maps to a variable, so for the example experiment, there are four layouts in the file: Drug, Concentration, Bacteria, and Killing.

A plater format file for the example experiment came with the package. Load plater (i.e. run library(plater)) and then run system.file("extdata", package = "plater"). Open the folder listed there and then open example-1.csv in a spreadsheet editor.

An abbreviated version of that file is shown below:

plater format example

The format is pretty simple:

You can use plater format with any standard plate size (6 to 1536 wells). Not every well has to be filled. If a well is blank in every layout in a file, it’s omitted. If it’s blank in some but not others, it’ll get NA where it’s blank.

While creating a file in plater format, it can be helpful to check whether you’re doing it right. For that purpose, you can pass the path of the file to check_plater_format(), which will check that the format is correct and diagnose any problems.

Step 2: Read in the data

Now that your file is set up, you’re ready to read in the data.

We will analyze this experiment two different ways to illustrate two common data analysis scenarios:

  1. Assuming the instrument gives back the killing data shaped like a plate, we’ll create one file with all four variables and read it in with read_plate().
  2. Assuming the instrument gives back tidy data (one-well-per-row), we’ll create two files–one with the data and one with the three variables–and then combine the files with add_plate().

Step 2: Read a single plater format file with read_plate()

Here is how it works. (Note that below we use system.file() here to get the file path of the example file, but for your own files you would specify the file path without using system.file()).

file_path <- system.file("extdata", "example-1.csv", package = "plater")
   
data <- read_plate(
      file = file_path,             # full path to the .csv file
      well_ids_column = "Wells"     # name to give column of well IDs (optional)
)
str(data)
#> Classes 'tbl_df', 'tbl' and 'data.frame':    96 obs. of  5 variables:
#>  $ Wells        : chr  "A01" "A02" "A03" "A04" ...
#>  $ Drug         : chr  "A" "A" "A" "A" ...
#>  $ Concentration: num  1.00e+02 2.00e+01 4.00 8.00e-01 1.60e-01 3.20e-02 6.40e-03 1.28e-03 2.56e-04 5.12e-05 ...
#>  $ Bacteria     : chr  "E. coli" "E. coli" "E. coli" "E. coli" ...
#>  $ Killing      : num  98 95 92 41 17 2 1.5 1.8 1 0.5 ...

head(data)
#>   Wells Drug Concentration Bacteria Killing
#> 1   A01    A       100.000  E. coli      98
#> 2   A02    A        20.000  E. coli      95
#> 3   A03    A         4.000  E. coli      92
#> 4   A04    A         0.800  E. coli      41
#> 5   A05    A         0.160  E. coli      17
#> 6   A06    A         0.032  E. coli       2

So what happened? read_plate() read in the plater format file you created and turned each layout into a column, using the name of the layout specified in the file. So you have four columns: Drug, Concentration, Bacteria, and Killing. It additionally creates a column named “Wells” with the well identifiers for each well. Now, each well is represented by a single row, with the values indicated in the file for each column.

Step 2 (again): Combine a one-well-per-row file and a plater format file with add_plate()

In the previous example, we assumed that the killing data was provided by the instrument in plate-shaped form, so it could just be pasted into the plater format file. Sometimes, though, you’ll get data back formatted with one well per row.

add_plate() is set up to help in this situation. You provide a tidy data frame including well IDs and then you provide a plater format file with the other information and add_plate() knits them together well-by-well. Here’s an example using the other two files installed along with plater.

file2A <- system.file("extdata", "example-2-part-A.csv", package = "plater")
data2 <- read.csv(file2A)

str(data2)
#> 'data.frame':    96 obs. of  2 variables:
#>  $ Wells  : chr  "A01" "A02" "A03" "A04" ...
#>  $ Killing: num  98 95 92 41 17 2 1.5 1.8 1 0.5 ...

head(data2)
#>   Wells Killing
#> 1   A01      98
#> 2   A02      95
#> 3   A03      92
#> 4   A04      41
#> 5   A05      17
#> 6   A06       2

meta <- system.file("extdata", "example-2-part-B.csv", package = "plater")
data2 <- add_plate(
      data = data2,               # data frame to add to 
      file = meta,                # full path to the .csv file
      well_ids_column = "Wells"   # name of column of well IDs in data frame
)

str(data2)
#> tibble [96 × 5] (S3: tbl_df/tbl/data.frame)
#>  $ Wells        : chr [1:96] "A01" "A02" "A03" "A04" ...
#>  $ Killing      : num [1:96] 98 95 92 41 17 2 1.5 1.8 1 0.5 ...
#>  $ Drug         : chr [1:96] "A" "A" "A" "A" ...
#>  $ Concentration: num [1:96] 1.00e+02 2.00e+01 4.00 8.00e-01 1.60e-01 3.20e-02 6.40e-03 1.28e-03 2.56e-04 5.12e-05 ...
#>  $ Bacteria     : chr [1:96] "E. coli" "E. coli" "E. coli" "E. coli" ...

head(data2)
#> # A tibble: 6 x 5
#>   Wells Killing Drug  Concentration Bacteria
#>   <chr>   <dbl> <chr>         <dbl> <chr>   
#> 1 A01        98 A           100     E. coli 
#> 2 A02        95 A            20     E. coli 
#> 3 A03        92 A             4     E. coli 
#> 4 A04        41 A             0.8   E. coli 
#> 5 A05        17 A             0.16  E. coli 
#> 6 A06         2 A             0.032 E. coli

add_plate then makes it easy to store data in a mix of formats, in some cases tidy and in some cases plate-shaped, which is the reality of many experiments.

Multiple plates

Say you were happy with the tests of you antibiotics, so you decided to do a second experiment, testing some other common pathogenic bacteria. Now you have data from two separate plates. Rather than handling them separately, you can combine them all into a common data frame with the read_plates() function.

Just like before, you create one plater file per plate, with all the information describing the experiment. In this case, you’ll have two files, one from each experiment. Then, just read them in with read_plates(). You can specify names for each plate, which will become a column in the output identifying which plate the well was on. By default it’ll use the file names.

# same file as above
file1 <- system.file("extdata", "example-1.csv", package = "plater")

# new file
file2 <- system.file("extdata", "more-bacteria.csv", package = "plater")

data <- read_plates(
   files = c(file1, file2),
   plate_names = c("Experiment 1", "Experiment 2"),
   well_ids_column = "Wells") # optional

str(data)
#> tibble [192 × 6] (S3: tbl_df/tbl/data.frame)
#>  $ Plate        : chr [1:192] "Experiment 1" "Experiment 1" "Experiment 1" "Experiment 1" ...
#>  $ Wells        : chr [1:192] "A01" "A02" "A03" "A04" ...
#>  $ Drug         : chr [1:192] "A" "A" "A" "A" ...
#>  $ Concentration: num [1:192] 1.00e+02 2.00e+01 4.00 8.00e-01 1.60e-01 3.20e-02 6.40e-03 1.28e-03 2.56e-04 5.12e-05 ...
#>  $ Bacteria     : chr [1:192] "E. coli" "E. coli" "E. coli" "E. coli" ...
#>  $ Killing      : num [1:192] 98 95 92 41 17 2 1.5 1.8 1 0.5 ...

head(data)
#> # A tibble: 6 x 6
#>   Plate        Wells Drug  Concentration Bacteria Killing
#>   <chr>        <chr> <chr>         <dbl> <chr>      <dbl>
#> 1 Experiment 1 A01   A           100     E. coli       98
#> 2 Experiment 1 A02   A            20     E. coli       95
#> 3 Experiment 1 A03   A             4     E. coli       92
#> 4 Experiment 1 A04   A             0.8   E. coli       41
#> 5 Experiment 1 A05   A             0.16  E. coli       17
#> 6 Experiment 1 A06   A             0.032 E. coli        2

Viewing plate-shaped data

Sometimes it’s useful to look back at the data in plate shape. Was there something weird about that one column? Was there contamination all in one corner of the plate?

For this, use view_plate() which takes a tidy data frame and displays columns from it as plate layouts.

view_plate(
  data = data2, 
  well_ids_column = "Wells", 
  columns_to_display = c("Concentration", "Killing")
)
#> $Concentration
#>     1  2 3   4    5     6      7       8        9       10        11 12
#> A 100 20 4 0.8 0.16 0.032 0.0064 0.00128 0.000256 5.12e-05 1.024e-05  0
#> B 100 20 4 0.8 0.16 0.032 0.0064 0.00128 0.000256 5.12e-05 1.024e-05  0
#> C 100 20 4 0.8 0.16 0.032 0.0064 0.00128 0.000256 5.12e-05 1.024e-05  0
#> D 100 20 4 0.8 0.16 0.032 0.0064 0.00128 0.000256 5.12e-05 1.024e-05  0
#> E 100 20 4 0.8 0.16 0.032 0.0064 0.00128 0.000256 5.12e-05 1.024e-05  0
#> F 100 20 4 0.8 0.16 0.032 0.0064 0.00128 0.000256 5.12e-05 1.024e-05  0
#> G 100 20 4 0.8 0.16 0.032 0.0064 0.00128 0.000256 5.12e-05 1.024e-05  0
#> H 100 20 4 0.8 0.16 0.032 0.0064 0.00128 0.000256 5.12e-05 1.024e-05  0
#> 
#> $Killing
#>     1   2   3   4   5   6   7   8   9  10  11  12
#> A  98  95  92  41  17   2 1.5 1.8   1 0.5 0.5 0.3
#> B  15   8   3 1.2 1.1 0.8 1.2 0.4 0.6 0.1 0.2 0.4
#> C  72  21   7 1.1 0.8 1.3 0.2 1.8   1 0.2 0.4 0.2
#> D 0.4 0.2 0.1 0.5 0.3 0.2 0.1 0.1 0.5 0.5 0.3 0.4
#> E  37   7   2 0.3 0.2 0.4 0.6 0.1   1 0.2 0.4 0.2
#> F  99  99  99  99  99  61   5 2.2 1.3 0.2 0.3 0.2
#> G  99  33   4 0.5 0.3 0.2 0.2 0.3 0.2 0.2 0.4 0.2
#> H  98  99  99  97  98  99  98  97  65  22   8 0.5