Casting data

In addition to making plots, R has tools to help you prepare tabular views of your data. This task can be viewed as a two-part process: first, selecting (aka subsetting) the data points that you want to present, and then arranging that data into the rows and columns that you need.

We have already seen how to use the subset() function to filter data. Let’s subset the LDLR data from the E1909 experiment.

subset(ablation, Measurement == "LDLR-ABLATION" & Experiment == "E1909")

To reshape this list of data points into a dataframe where there is one row per time point and one column per cell type, we can use the dcast() function from the reshape2 package (this package is installed with ggplot2 but must be loaded to use the functions).

library(reshape2)
dcast(subset(ablation,
             Measurement == "LDLR-ABLATION" & Experiment == "E1909"),
       Time ~ CellType,
       value.var = "Score")

Note that all of the experimentally measured values in this table come from the original Score column; this is indicated by the value.var option in the above command. The values in the Time column, and the column names, are referred to as identifying variables. In other words, if you are looking up a specific experimentally measured value, you would need to specify the identifying variables (Time and CellType) in order to find it.

When using dcast(), the factors on the RHS of the tilde will be the rows, and the factors on the LHS will be the columns. You can specify more than one identifying variable on either side of the tilde.

Subset the E1909 experiment data from the ablation dataset, and assign it to a variable called e1909. Reshape this data such that you have one row per unique combination of Measurement and Time, and one column per CellType.

Note that if you can also use more than one identifier in the column attribute, the column names are amalgamations of the originating identifiers.

Introduction

Data Structures

Plotting

Data wrangling

Exercise

Casting data

Instructions