Skip to content

Factors and Coercion

Introducing factors

Factors are another major data structure that is important to know about. Factors can be thought of as vectors which are specialized for categorical data. Given R's specialization for statistics, this make sense since categorial and continuous variables are usually treated differently. Sometimes you may want to have data treated as a factor, but in other cases, this may be undesirable. Let's see the value of treating some of which are categorical in nature as factors. Let's take a look at just the alternate alleles, continuing on from our variants dataframe example.

r

# Recreate the variants dataframe and subset it, if you do not already have these objects in your environment:

variants <- read.csv("https://raw.githubusercontent.com/GenomicsAotearoa/Introduction-to-R/refs/heads/main/combined_tidy_vcf.csv")
subset_variants <- variants[, c(1:3, 6)]

r

# Extract the "ALT" column to a new object

alt_alleles <- subset_variants$ALT

Let's look at the first few items in our factor using head():

r

head(alt_alleles)
Output 
[1] "G"         "T"         "T"         "CTTTTTTTT" "CCGCGC"    "T"

There are 801 alleles (one for each row). To simplify, lets look at just the single-nucleotide alleles (SNPs). We can use some of the vector indexing skills from the last episode.

r

alt_snps <- c(alt_alleles[alt_alleles=="A"],
              alt_alleles[alt_alleles=="T"],
              alt_alleles[alt_alleles=="G"],
              alt_alleles[alt_alleles=="C"])

This leaves us with a vector of the 701 alternative alleles which were single nucleotides. Right now, they are being treated a characters, but we could treat them as categories of SNP. Doing this will enable some nice features. For example, we can try to generate a plot of this character vector as it is right now:

r

plot(alt_snps)
Output 
Error in plot.window(...) : need finite 'ylim' values
In addition: Warning messages:
1: In xy.coords(x, y, xlabel, ylabel, log) : NAs introduced by coercion
2: In min(x) : no non-missing arguments to min; returning Inf
3: In max(x) : no non-missing arguments to max; returning -Inf

Whoops! Though the plot() function will do its best to give us a quick plot, it is unable to do so here. One way to fix this it to tell R to treat the SNPs as categories (i.e., a factor vector); we will create a new object to avoid confusion using the factor() function:

r

factor_snps <- factor(alt_snps)

Let's learn a little more about this new type of vector:

r

str(factor_snps)
Output 
 Factor w/ 4 levels "A","C","G","T": 1 1 1 1 1 1 1 1 1 1 ...

What we get back are the categories ("A","C","G","T") in our factor; these are called "levels". Levels are the different categories contained in a factor. By default, R will organize the levels in a factor in alphabetical order. So the first level in this factor is "A".

For the sake of efficiency, R stores the content of a factor as a vector of integers, which an integer is assigned to each of the possible levels. Recall levels are assigned in alphabetical order. In this case, the first item in our factor_snps object is "A", which happens to be the 1st level of our factor, ordered alphabetically. This explains the sequence of "1"s (Factor w/ 4 levels"A","C","G","T": 1 1 1 1 1 1 1 1 1 1 ...), since "A" is the first level and the first few items in our factor are all "A"s.

We can see how many items in our vector fall into each category:

r

summary(factor_snps)
Output 
  A   C   G   T 
211 139 154 203

As you can imagine, this is already useful when you want to generate a tally.

Treating objects as categories without changing their mode

You don't have to make an object a factor to get the benefits of treating an object as a factor. See what happens when you use the as.factor() function on factor_snps. To generate a tally, you can sometimes also use the table() function; though sometimes you may need to combine both (i.e., table(as.factor(object)))

Plotting and ordering factors

One of the most common uses for factors will be when you plot categorical values. For example, suppose we want to know how many of our variants had each possible SNP we could generate a plot:

r

plot(factor_snps)
Output

images

This isn't a particularly pretty example of a plot but it works. We'll be learning much more about creating nice, publication-quality graphics later in this workshop.

If you recall, factors are ordered alphabetically. That might make sense, but categories (e.g., "red", "blue", "green") often do not have an intrinsic order. What if we wanted to order our plot according to the numerical value (i.e., in descending order of SNP frequency)? We can enforce an order on our factors:

r

ordered_factor_snps <- factor(
  factor_snps, 
  levels = names(sort(table(factor_snps)))
)

Let's deconstruct this from the inside out (you can try each of these commands to see why this works):

  1. We create a table of factor_snps to get the frequency of each SNP: table(factor_snps)
  2. We sort this table: sort(table(factor_snps)); use the decreasing = parameter for this function if you wanted to change from the default of FALSE
  3. Using the names() function gives us just the character names of the table sorted by frequencies:names(sort(table(factor_snps)))
  4. The factor function is what allows us to create a factor. We give it the factor_snps object as input, and use the levels= parameter to enforce the ordering of the levels.

Now we see our plot has be reordered:

r

plot(ordered_factor_snps)
Output

images

Factors come in handy in many places when using R. Even using more sophisticated plotting packages such as ggplot2 will sometimes require you to understand how to manipulate factors.

Coercing values

Sometimes, it is possible that R will misinterpret the type of data represented in a data frame, or store that data in a mode which prevents you from operating on the data the way you wish. For example, a long list of gene names isn't usually thought of as a categorical variable, the way that your experimental condition (e.g., control, treatment) might be. More importantly, some R packages you use to analyze your data may expect characters as input, not factors. At other times (such as plotting or some statistical analyses) a factor may be more appropriate. Ultimately, you should know how to change the mode of an object.

First, its very important to recognize that coercion happens in R all the time. This can be a good thing when R gets it right, or a bad thing when the result is not what you expect. Consider:

r

snp_chromosomes <- c('3', '11', 'X', '6')
typeof(snp_chromosomes)
Output 
[1] "character"

Although there are several numbers in our vector, they are all in quotes, so we have explicitly told R to consider them as characters. However, even if we removed the quotes from the numbers, R would coerce everything into a character:

r

snp_chromosomes_2 <- c(3, 11, 'X', 6)
typeof(snp_chromosomes_2)
snp_chromosomes_2[1]
Output 
[1] "character"
[1] "3"

We can use the as. functions to explicitly coerce values from one form into another. Consider the following vector of characters, which all happen to be valid numbers:

r

snp_positions_2 <- c("8762685", "66560624", "67545785", "154039662")
typeof(snp_positions_2)
snp_positions_2[1]
Output 
[1] "character"
[1] "8762685"

Now we can coerce snp_positions_2 into a numeric mode using as.numeric():

r

snp_positions_2 <- as.numeric(snp_positions_2)
typeof(snp_positions_2)
snp_positions_2[1]
Output 
[1] "double"
[1] 8762685

Sometimes coercion is straight forward, but what would happen if we tried using as.numeric() on snp_chromosomes_2

r

snp_chromosomes_2 <- as.numeric(snp_chromosomes_2)
Output 
Warning message:
NAs introduced by coercion

If we check, we will see that an NA value (R's default value for missing data) has been introduced.

r

snp_chromosomes_2
Output 
[1]  3 11 NA  6

Trouble can really start when we try to coerce a factor. For example, when we try to coerce the factor_snps into a numeric mode look at the result:

r

as.numeric(factor_snps)
Output 
  [1] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
 [28] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
 [55] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
 [82] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
[109] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
[136] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
[163] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
[190] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 4 4 4 4 4
[217] 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
[244] 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
[271] 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
[298] 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
[325] 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
[352] 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
[379] 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
[406] 4 4 4 4 4 4 4 4 4 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
[433] 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
[460] 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
[487] 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
[514] 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
[541] 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
[568] 3 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
[595] 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
[622] 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
[649] 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
[676] 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
[703] 2 2 2 2 2

Strangely, it works! Almost. Instead of giving an error message, R returns numeric values, which in this case are the integers assigned to the levels in this factor. This kind of behavior can lead to hard-to-find bugs, for example when we do have numbers in a factor, and we get numbers from a coercion. If we don't look carefully, we may not notice a problem.

If you need to coerce an entire column you can overwrite it using an expression like this one:

r

# Make the 'sample_id' column a factor type column
variants$sample_id <- as.factor(variants$sample_id)

# check the structure of the column
str(variants$sample_id)
Output 
Factor w/ 3 levels "SRR2584863","SRR2584866",..: 1 1 1 1 1 1 1 1 1 1 ...

Lesson summary: Data coercion

Lets summarize this section on coercion with a few take home messages.

  • When you explicitly coerce one data type into another (this is known as explicit coercion), be careful to check the result. Ideally, you should try to see if its possible to avoid steps in your analysis that force you to coerce.
  • R will sometimes coerce without you asking for it. This is called (appropriately) implicit coercion. For example when we tried to create a vector with multiple data types, R chose one type through implicit coercion.
  • Check the structure (str()) of your data before working with them!

StringsAsFactors

Prior to R 4.0 when importing a data frame using any one of the read.table() functions such as read.csv() , the argument StringsAsFactors was by default set to true TRUE. Setting it to FALSE will treat any non-numeric column to a character type. read.csv() documentation, you will also see you can explicitly type your columns using the colClasses argument. Other R packages (such as the Tidyverse readr) don't have this particular conversion issue, but many packages will still try to guess a data type.