2. What School / Dept? n=55

3. What is the primary research question you work on?

4. Why you want to participate?

5. School of Biotechnology
Data for life:
An Introduction to R
Dr. Tim Downing
School of Biotechnology
See Resources and References
#dataforlife 5

7. Social Networking Sites
metabolite data
ion channels + prostate cancer
blockchain technology
Cybersecurity and parenting
gene copies and visualization
cycling ability
marketing + market development
NLP for morphologically rich languages
operational efficiencies + data analytics
e-commerce
wearable fitness technology
proteomics and phosphoproteomic analysis
how to predict future
Pancreatic Cancer Research
drugs interaction with biological membrane
Intelligent Power Systems
violent online political extremism
literacy attainment for deaf and hard of hearing children
Trends in combat sports
biomanufacturability of CHO cells
Collaboration and interorganisational relationships
capturing carbon dioxide and using it as a raw material
Predicting the sustainability and resale value of a car

8. Who owns the data? Example: NHS March 2018: sequence genomes
100,000 Genomes Project -> 32k
Genetics Expert Network for Enterprises (GENE)
13 private companies
AbbVie
Alexion Pharmaceuticals
AstraZeneca
Biogen
Dimension Therapeutics
GSK
Helomics
Roche
Takeda
UCB*
March 2018: sequence genomes
Jan 2019: NHS own
genomes as database
Who owns the data?

9. Example: NHS
Google?
99.7% "Deriving genomic diagnoses without revealing patient genomes"
Report at

11. Code not spreadsheets!
data in formulas! =(3.6946*10^-6)/'Old snails'!J26
ref: Jenny Bryan

12. ref: Jenny Bryan

14. Code not spreadsheets!

15. Code not spreadsheets!
Do something once, you may not need a script. But, do it hundreds or thousands of times, you need one(s)
Share what you did with a script
This may be complicated by tool version numbers
ref: BF Francis Ouellette

16. Mistakes matter!
Paper: "Genomic surveillance elucidates Ebola virus origin and transmission during the 2014 outbreak mutation rate and genotype variation of Ebola virus from Mali case sequences" DOI: /science.aaa5646
Wait, WTF? Comment on “Mutation rate and genotype variation of Ebola virus from Mali case sequences” DOI: /science.aaf3823
Oops OMG rly sry Response to Comment on “Mutation rate and genotype variation of Ebola virus from Mali case sequences” DOI: /science.aaf4561

17. Fig. 2 Maximum likelihood tree of the 106 sequences analyzed by Hoenen et al. (left side) initially (1) with lines linking to the correctly labeled sequences after the erratum (3) (right side).
Maximum likelihood tree of the 106 sequences analyzed by Hoenen et al. (left side) initially (1) with lines linking to the correctly labeled sequences after the erratum (3) (right side). Lines of the same color represent multiple samples taken from the same patient, which in most cases have identical sequences. These correctly group together on the right but do not in many cases on the left.
Rambaut et al. Science 2016

18. Murphy's Law / Reviewer 3 - How exactly did you define your variables?
- You had 14,327 records at the start, but the analysis is based on 14,225. What happened to the other 122?
- Why did you exclude the measurements from Sensor D, but only b/w 2pm and 4pm on the 18th day?
- You realise that you've normalised the Loop grades incorrectly, the average is 62%, not 153%
- Figure 4 is great, but can you to make the axis labels one louder and move the legend to the top left hand corner beside the dancing leprechaun? 18

19. Code not spreadsheets! For Every Result, Track How It Was Produced
Avoid Manual Data Manipulation
Archive Exact Versions of All Programs Used
Version Control All Scripts
Record All Intermediate Results in Standard Formats
Note Random Seeds/Numbers
Always Store Raw Data behind Plots
Allow Layers of Increasing Detail to Be Inspected
Connect Textual Statements to Underlying Results
Provide Public Access to Scripts, Runs and Results

20. What is R / RStudio?
- An environment for statistical analysis and graphics
- Supports almost every statistical method
- The environment of choice for developing new methods
It's for anyone who has to analyse data
- from A to Z
- agriculture, astrophysics, climatology,
- ecology and environmental science, econometrics,
- electrical engineering, finance, genetics,
- genomics, geography, psychology, public health,
- social sciences, zoology! 20

21. R Course Aims - Introduce you to R
- Get you to use R yourself on your own data
- Work in groups to talk about using R in your projects
- Show you how to learn what you need to know about R 21

22. R Course Learning Outcomes
At the end of this session you will be able to:
Start R Studio and create a Rmd (R Markdown) file
Understand what a Rmd file is
Load a simple data file
Understand how R uses dataframes and vectors
Use the R help system
Examine a dataframe
Prepare simple tables
Locate and perform some stats on your data
(Produce good quality graphs, and save these)
(Prepare a simple function) 22

23. R code terms Rmd file – useful text file of your code and comments
vector – a list of numbers
variable – something we define, eg x
function – R code that does something specific, eg mean(x)
arguments – data required by a function, eg x above
library – package of R functions that does something specific loaded using the library() function
environment - the space you are working in, and the set of things that can be found there (eg Windows, Mac, etc)
CRAN – website from with libraries to meet your needs

24. Managing projects in R Many options: I prefer Rmd files (R Markdown)
Record your commands in your notebook (I use Evernote) and in your Rmd file
Using GUI (eg Excel) or websites is asking for trouble
FAIR etc 24

25. Managing projects in R - Keep your work organised
- Keep your original data
- How did you get from there to here?
- Record all analyses, both the ones you published and the ones you didn't
- Start with the original data and re-trace your steps
- Never overwrite the original data, disk space = cheap 25

26. Cleaning data / Data QC - 90% of the work - Errors in the data
- Fix these
- Or omit these, and document why
- Create new variables, and recode old ones, and all of this is done live in R
- Output a dataset ready for further analysis 26

27. Actually doing the thing you wanted to do in the first place after all that annoying checking that took forever omfg
10% of the work
- Prepare tables and graphs from the data
- Carry out statistical tests
- Start with simple models & complex ones if needed
All with R
(Perl, Python, Java, Jython, C, C++, Ruby, Fortran, ...) 27

28. 28

29. Introduction to R After this section you should be able to:
- Read data in and output data from R
- Manipulate data in R
- Use some basic R functions (eg statistics and plotting)
You may be able to write basic R computer programs
Use it or lose it
There is a lot to R beyond this course – more available in books, tutorials, online forums, R demos, help pages etc

30. Using your own laptop Website
Download R from the R Project website install it and start it
Download RStudio from the RStudio website install it and start it
Explore "A very short introduction to R" from CRAN

31. R Course Data
Download these 31

32. Oh no. I can't immediately install R & RStudio
Oh no! I can't immediately install R & RStudio! Try the online versions:
(Good)
(the first part is free, which is enough)

33. What RStudio looks like
Environment
Scripting
Plots
Console

34. Running commands Click anywhere in the line with your cursor
Ctrl + Return to run it or click on the Run button 34

35. Introduction to R Help( ) > help(topic) help() is an R function
R functions take arguments (info inside the function: they which go in between brackets, separated by commas)
‘help’ function => displays info on R documentation files

36. Introduction to R R as calculator
R will evaluate basic calculations which you type into the console (input window)
Type in “R”
> 10+1
[1] 11
> 10*2
[1] 20
> (10*2)+1 use brackets to clarify precedence
[1] 21

37. Introduction to R Assigning variables > x <- 1.5
> y <- 2.6
> x check what x and y were assigned as
[1] 1.5
> y
[1] 2.6

38. Introduction to R Assigning variables > x <- 1.5
> y <- 2.6
R uses standard operators / *
a**b a to the power of b
a %% b a modulo b
(get the remainder)
> (x/(y + 20/15))**18 R can do complex operations
[1] e-08

39. Introduction to R
Generally can obtain the number (or other value) stored in any variable by typing the name of the variable, followed by enter; or print (variable) or show(variable)
> x
[1] 1.5
Wait, what did I assign x as again … ?
> print (x)
> show(x)

40. Introduction to R
Now try the questions!

41. Introduction to R Vectors
Vectors = a single column/row of numbers in a spreadsheet
Create a vector using the c() function (concatenate):
x <- c()
eg x <- c(1,2,4,8) creates a column of the numbers 1,2,4,8

42. Introduction to R For simple operations (+ - * /) on vectors x and y:
If x and y have same number of entries:
normal operations on numbers in vector entry by entry
Else x and y have different number of entries:
for vector with fewest entries
x <- c(1,2,3) # 3 elements in x
y <- c(1,2,3,4) # 4 elements in y
z <- x*y # This causes R to panic, it can't handle this
# merged elements must contain same #elements

43. Introduction to R Vectors x y v > x <- c(1,2,3,4)
> y <- c(55,66,77,88)
> z <- c(10,11,12)
> v <- x*y multiple commands on one line
> v
[1]
> v <- x/y
[1]

44. Introduction to R We can do standard operations on vectors x y v
[1]
> v + x
[1]
> v + x*y
[1]
> v <- c(x,y) create a longer vector v
> v
[1]

45. Introduction to R
Some more useful ways of creating columns of numbers (vectors)
The seq function
seq(1,10,1) = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
seq(1,4,0.5) = 1, 1.5, 2, 2.5, 3, 3.5, 4
x:y
1:10 = 1,2,3,4,5,6,7,8,9,10
The rep function
rep(2,4) = 2, 2, 2, 2

46. Introduction to R > seq(1,10) [1] 1 2 3 4 5 6 7 8 9 10
[1]
> seq(1,21,2.5)
[1]
> 1:9
[1]
> rep(2,3)
[1] 2 2 2
> rep(3,5)
[1]

47. Introduction to R Get specific parts of a vector:
Example: if x = vector 1, 2, 5 then x[1] = 1, x[2] = 2, x[3] = 5
> x <- c(1,2,5)
> x
[1] 1 2 5
> x[1]
[1] 1
> x[2]
[1] 2
> x[3]
[1] 5

48. Introduction to R Use square brackets to refer to 2+ cells of a vector
Examples
x <- c(1,2,5)
y <- c(1,5)
c(y, y) 1, 5, 1, 5
x[1:2] 1, 2
x[2:3] 2, 5
x[c(1,3)] 1, 5

49. Introduction to R More useful stuff...
ls() # lists all variables/objects you made
rm(list = ls()) # removes ALL objects!
A vector index can be a condition as well as a number
Example x <- c(1, 0, 8, 12, 0)
Get positive elements of x x[x>0]
[1]

50. Introduction to R vectors can consist of strings of letters
x <- c (“purple”, “green”, “blue”)
A vector never consists of numbers AND strings
If numbers and strings in a vector, R considers every entry in vector to be a string (ie not viable for maths operations etc)

51. Set your working directory
On the top menu, go to "Session"
Then "Set Working Directory"
Then "Choose Directory" - select the location you want
or getwd()
setwd("C:/Users/DowningT/myfolder/")
Now: Save your commands
(eg a Notepad text file, Evernote, R studio, etc)

52. Introduction to R Data frames (and the function data.frame( ) )
Ordered list (entries correspond to one another)
Each element of the list has the same length
h <- c (150, 170, 168, 179, 130)
w <- c (65, 70, 72, 80, 51)
patient_data <- data.frame (weights=w, heights=h)

53. Introduction to R
Accessing a particular column, cell or row of a data frame
Accessing a cell
patient_data [i,j] (jth cell in the ith column)
Accessing a row
patient_data [i,] (row i)
Accessing a column
patient_data [,i] (column i)

54. Introduction to R
Factors = data type with info about observations original group
heights <- c(1.7,1.95,1.63,1.54,1.29)
Make new vector fac_heights to record nationalities:
fac_heights <- factor(c(“GB”, “IR”, “GB”, “GB”, “IR”))
Useful for statistical tests b/w groups

55. Introduction to R Finding out about data object
mode(): tells you storage
mode of object
class(): info on object’s class
often determines how object is handled by a function
You can also set object’s mode, attributes or class using above functions.
mode(x) <- "numeric"
mode(y) <- "character"

56. Data input and output R comes with several pre-packaged datasets
You can access these datasets with the data function
data() gets you a list of all the datasets
data(Titanic) loads a dataset about passengers on the Titanic (for example, others include data(tea) etc)
summary(Titanic) provides some summary information about the dataset Titanic
attributes(Titanic) provides some more information
Typing the dataset name on its own (followed by Enter) will display the data

57. Data input and output R comes with several pre-packaged datasets
You can access these datasets with the data function
eg 1990 Davis PMID
"Body image and weight preoccupation: A comparison between exercising and non-exercising women"
View(Davis)
head(Davis)
str(Davis)
glimpse(Davis)
summary(Davis)

58. Data input and output Davis table(Davis$weight) table(Davis$height)
data.frame(Davis$height,Davis$weight)
# create a dataframe from heights and weights only
data.frame(Davis$height,Davis$weight)[1:5,]
look at dataframe for heights and weights - samples 1-5

59. Data input and output
Can use function read.table() to read in dataframe:
q <- read.table(file.choose(), sep=“\t”, header=T)

60. CSV data First line = names of variables, separated by commas
Variables = proper numbers or plain text - no spaces,
funny characters or punctuation
Data = mix of numbers and grouping variables
= time sequence is ok
= dates and times are not
Missing data = represented by the two letters NA and
nothing else - no dashes, no 999, 77, 88 or anything else

61. CSV data

62. Data input and output
Writing data to a file (the write() and write.table() functions)
Write list out to current directory
write.table (q, file = “filename”, ncol = 2)
for vector, ncol specifies #columns in output
For a data frame
(many optional
arguments ... )

63. For loops Writing your own functions in R Syntax: let x be a vector
for ( i in x )
command
eg try for(i in 1:10) {
print(i) }
Writing your own functions in R

64. For loops Writing your own functions in R
Example: add the elements of a vector x <- c (1, 2, 8)
sum <- 0
for (i in x)
sum <- sum + i
sum = 11
sum() does this for you
Writing your own functions in R

65. Conditional statements
If statement execution
Syntax:
if (condition){
command } OR
x <- 11
if(x>2){ print (x) }
Writing your own functions in R

66. Conditional execution
>, <, <=, >=, !=, == are all used to compare numbers
= means assign
These can also be used in conditional indexing (see earlier)
You can combine conditions with the & or | sign
AND OR
Writing your own functions in R

67. Combined loop & conditional execution
Example:
add positive elements of vector
x <- c (1, -2, 8)
pos_sum <- 0
for (i in x)
if (i > 0)
pos_sum <- pos_sum + i
pos_sum
This is for illustration – there are quicker ways of doing this
Writing your own functions in R

68. Blocks Writing your own functions in R
Sometimes you need to execute several commands in a loop or after testing a condition
R will group together all commands within { }
x <- 11
if(x>2){ print (x) }
Writing your own functions in R

69. Blocks Writing your own functions in R
Add up all the numbers between 1 and 10 and multiply by the sum of the numbers between 5 and 10
for ( i in 1:10) {
m <- m + i 1, 3, 6, …
if ( i > 5 ) { k <- k + i } ,13,… }
s <- m * k s[3]=m[1]*k[3], s[4]=m[2]*k[4]
Writing your own functions in R

70. Writing your own functions in R
Making your own function is possible:
my_function <- function ( ... ) {}
fix ( my_function )
Note: comment your code (anything after the # symbol is ignored by R – this is a place to put ‘notes to self’)
for ( i in 1:10) { # for each number 1 to 10
m <- m + i # m is assigned as itself + i
if ( i > 5 ) { k <- k + i } # if i>5, k is assigned as itself + i
} # end for
s <- m * k # s is m by k
Writing your own functions in R

71. Scientific method in biostatistics
1. Defining problems:
What type of data is being tested?
2. Assumptions of each model:
Is this valid for this data?
Why has this particular model been used?
3. Testing hypothesis predictions:
If X is true, then what follows?
4. Boundary conditions:
When does it fail or not work?

72. Getting your data into R
Download -> CSV -> data.frame -> data cleaning
At the end of this session, you will be able to:
Upload, download, create or import data into R
Manipulate large datasets as tables in R
Explore datasets using multiple approaches
Test for missing, partial or inconsistent data
Summarise and compare datasets with R

73. Dplyr -> Data cleaning + EDA
=> manipulating data:
Davis %>% filter(height<quantile(height,0.5))
# subset rows of smallest 50%
%>% arrange(desc(height))
# sort rows of smallest 50%
%>% select(sex,weight,repwt)
# select columns we want
%>% mutate(weight_diff=(weight-repwt))
# create new variables, eg "weight_diff"
%>% group_by(sex)
%>% summarise(mean=mean(weight))
# summarise details of interest