Every value in R has a type, and the type decides what you can do with it: you can add two read counts, but not a read count and a gene name. In this lab you will practise the lecture’s topics using small examples from the lab bench. By the end you should be able to:
Inf, NaN and NA,The four data types you will use every day:
| Type | What it holds | Lab example |
|---|---|---|
| double | any number | Ct value 24.1, temperature 37.5 |
| integer | whole numbers | read count 1523L |
| character | text | gene "BRCA1", sample "S01" |
| logical | TRUE or FALSE |
treated? TRUE |
Numbers are stored as doubles by default, even whole numbers like 1523. Add L to get an integer: 1523L.
In all exercises in this course, first try to work out what the result should be, then run the code to check. If the output surprises you, make sure you understand why before you move on. If you get stuck, click the key to reveal example code and output, or ask a TA or the person next to you.
There are often several equally good solutions.
We recommend writing all your code in your R Markdown learning journal (from the first lab). By the end of the course it will be your own R tutorial, with your notes and solutions.
Open RStudio and open your r-course Project (File → Open Project…), so that R works in the right folder. You do not need setwd() when you use a Project.
List all objects in your current R session:
ls()
## [1] "address" "answer_box" "asst"
## [4] "blank" "cnames" "colonies"
## [7] "color_primary" "color_text" "combined"
## [10] "crashes" "dfr" "dilution"
## [13] "doctor" "drinks" "eng_output"
## [16] "fa" "fa1" "fa2"
## [19] "gapminder" "gene" "glucose"
## [22] "knit_walkthrough" "lego" "linkoping"
## [25] "loc" "lund" "page_dir"
## [28] "patients" "penguins" "plate"
## [31] "plate_corrected" "plate_test" "r"
## [34] "readings" "res" "s"
## [37] "sharks" "stackoverflow" "umea"
## [40] "uppsala" "uppsala2" "uppsala3"
## [43] "videogames" "volume_ml" "walkthrough_block"
## [46] "wt_colonies" "wt_imposter" "wt_mice"
## [49] "wt_primers"
If you want a clean start, the safest option is Session → Restart R. You can also remove all objects with:
rm(list = ls())
This deletes everything in your environment. It uses functions we have not covered yet; it will make sense after the next lectures.
TRUE). Store each value in a well-named variable and check its type with typeof(). Is the number of pups an integer? If not, make it one.
mouse_id <- "M07"
weight_g <- 23.4
n_pups <- 6
knockout <- TRUE
typeof(mouse_id)
typeof(weight_g)
typeof(n_pups) # "double": plain numbers are doubles
typeof(knockout)
n_pups <- 6L # the L suffix makes an integer
typeof(n_pups)
## [1] "character"
## [1] "double"
## [1] "double"
## [1] "logical"
## [1] "integer"
A colleague stored the weight as "23.4" (in quotes). What happens if you add 1.5 g to it? How do you fix it?
weight_text <- "23.4"
weight_text + 1.5 # error: you cannot do arithmetic on text
## Error in weight_text + 1.5: non-numeric argument to binary operator
as.numeric(weight_text) + 1.5 # convert first
## [1] 24.9
stock_conc <- 250
target_conc <- 10
final_volume <- 20
stock_volume <- target_conc * final_volume / stock_conc
water_volume <- final_volume - stock_volume
stock_volume
water_volume
## [1] 0.8
## [1] 19.2
as.integer(), round(), floor() and ceiling() on 250 / 96. Which one gives the right answer here, and why are the others wrong?
250 / 96
as.integer(250 / 96) # 2: cuts off decimals, 58 samples have no plate!
round(250 / 96) # 3, but only by luck: try 200 / 96
floor(250 / 96) # 2: same problem as as.integer()
ceiling(250 / 96) # 3: always rounds up, which is what we need
## [1] 2.604167
## [1] 2
## [1] 3
## [1] 2
## [1] 3
"OVER") and one was left empty:absorbance <- c("0.84", "1.02", "OVER", "0.77", "")
What is the type of absorbance? Convert it to numbers and count how many values became NA. Then calculate the mean absorbance of the wells that were measured.
typeof(absorbance)
abs_num <- as.numeric(absorbance) # warning: NAs introduced by coercion
abs_num
sum(is.na(abs_num)) # 2
mean(abs_num, na.rm = TRUE)
## [1] "character"
## [1] 0.84 1.02 NA 0.77 NA
## [1] 2
## [1] 0.8766667
Write a sentence in your journal: "OVER" and "" both became NA, but do they mean the same thing? Would dropping the "OVER" well bias your result?
typeof():c(TRUE, 2L)
c("37", 5)
c(NA, 1L)
c(NA, "control")
typeof(c(TRUE, 2L)) # integer
typeof(c("37", 5)) # character: the number 5 becomes the text "5"
typeof(c(NA, 1L)) # integer: NA fits into any type
typeof(c(NA, "control")) # character
## [1] "integer"
## [1] "character"
## [1] "integer"
## [1] "character"
| Well | Infected cells | Total cells |
|---|---|---|
| A1 | 12 | 40 |
| A2 | 0 | 35 |
| A3 | 0 | 0 |
Then calculate the log10 of these viral titres: 1,000,000, 1,000 and 0.
12 / 40 # 0.3
0 / 35 # 0
0 / 0 # NaN: an empty well has no infection rate
log10(1e6) # 6
log10(1e3) # 3
log10(0) # -Inf
## [1] 0.3
## [1] 0
## [1] NaN
## [1] 6
## [1] 3
## [1] -Inf
heights <- c(12.5, NA, 9.8, 15.1, 11.0, NA)
How many values are missing? What is the tallest plant? How many plants are taller than 10 cm? Why do you need na.rm = TRUE in all of them?
sum(is.na(heights))
max(heights) # NA
max(heights, na.rm = TRUE)
sum(heights > 10) # NA: we don't know about the missing plants
sum(heights > 10, na.rm = TRUE)
## [1] 2
## [1] NA
## [1] 15.1
## [1] NA
## [1] 3
p = 0.65 and q = 1 - p, the Hardy–Weinberg genotype frequencies are p², 2pq and q². Calculate them, then check that they add up to 1, first with == and then with all.equal(). What happened?
p <- 0.65
q <- 1 - p
p^2
2 * p * q
q^2
p^2 + 2 * p * q + q^2 == 1 # FALSE: rounding errors in decimals
isTRUE(all.equal(p^2 + 2 * p * q + q^2, 1)) # TRUE
## [1] 0.4225
## [1] 0.455
## [1] 0.1225
## [1] FALSE
## [1] TRUE
sprintf() or round() to report it as a whole number.
start_cells <- 1e5
doubling_time <- 20
hours <- 72
cells <- start_cells * 2^(hours / doubling_time)
cells
round(cells)
## [1] 1212573
## [1] 1212573
age <- 52
bmi <- 27.4
diabetic <- TRUE
age >= 18 & age <= 65 & (bmi > 30 | diabetic) # TRUE
# A 70-year-old diabetic patient should NOT be included:
age <- 70
age >= 18 & age <= 65 & (bmi > 30 | diabetic) # FALSE: correct
age >= 18 & age <= 65 & bmi > 30 | diabetic # TRUE: wrong! & is evaluated before |,
# so every diabetic patient passes
## [1] TRUE
## [1] FALSE
## [1] TRUE
mean() of TRUE/FALSE values give you?quality <- c(35, 28, 40, 31, 19, 38, 33, 25)
quality >= 30
sum(quality >= 30) # 5 reads
mean(quality >= 30) # 0.625: the fraction of TRUE values
## [1] TRUE FALSE TRUE TRUE FALSE TRUE TRUE FALSE
## [1] 5
## [1] 0.625
sort(c("S1", "S2", "S10", "S11", "S3"))
"Control" == "control"
sqrt(2)^2 == 2
sort(c("S1", "S2", "S10", "S11", "S3")) # "S10" and "S11" come before "S2"
"Control" == "control" # FALSE: comparisons are case-sensitive
sqrt(2)^2 == 2 # FALSE: decimals are not stored exactly
isTRUE(all.equal(sqrt(2)^2, 2)) # TRUE
## [1] "S1" "S10" "S11" "S2" "S3"
## [1] FALSE
## [1] FALSE
## [1] TRUE
primer <- "ACGTTGCAGGCT"
?toupper)."ACG"? (Extract the first three bases and compare.)T with U. Try sub() first, then gsub(). What is the difference?
nchar(primer)
tolower(primer)
substr(primer, start = 1, stop = 3) == "ACG"
sub("T", "U", primer) # replaces only the first T
gsub("T", "U", primer) # replaces every T
## [1] 12
## [1] "acgttgcaggct"
## [1] TRUE
## [1] "ACGUTGCAGGCT"
## [1] "ACGUUGCAGGCU"
paste(), paste0() and sub():"A1", "A2", …, "A12","mouse_1.fastq", "mouse_2.fastq", "mouse_3.fastq","S3" from the file name "S3_R1.fastq.gz".
paste0("A", 1:12)
paste0("mouse_", 1:3, ".fastq")
sub("_R1.fastq.gz", "", "S3_R1.fastq.gz")
## [1] "A1" "A2" "A3" "A4" "A5" "A6" "A7" "A8" "A9" "A10" "A11" "A12"
## [1] "mouse_1.fastq" "mouse_2.fastq" "mouse_3.fastq"
## [1] "S3"
sprintf() and your variables from exercise 2, print this pipetting instruction:Mix 0.80 uL stock with 19.20 uL water
sprintf("Mix %.2f uL stock with %.2f uL water", stock_volume, water_volume)
## [1] "Mix 0.80 uL stock with 19.20 uL water"