Introduction to R, Day 1:

“Getting oriented”
Michael Chimenti and Diana Kolbe
Dec 5 & 6 2019
Iowa Institute of Human Genetics

Slides adapted from HPC Bio at Univ. of Illinois:

https://wiki.illinois.edu/wiki/pages/viewpage.action?pageId=705021292

Distributed under a CC Attribution ShareAlike v4.0 license (adapted work):

https://creativecommons.org/licenses/by-sa/4.0/
Learning objectives:

1. Be able to describe what R is and how the programming

environment works.

2. Be able to use RStudio to install add-on packages on your

own computer.

3. Be able to read, understand and write simple R code

4. Know how to get help.

5. Be able to describe the differences between R, RStudio and

Bioconductor
What is R? (www.r-project.org)

• "... a system for statistical computation and graphics" consisting of:

1. A simple and effective programming language

2. A run-time environment with graphics

• Many statistical procedures available for R; currently 15,111 additional

add-on packages in CRAN

• Completely free, open source, and available for Windows, Unix/Linux,

and Mac
What is Bioconductor? (www.bioconductor.org)
• “… open source, open development software project to provide tools
for the analysis and comprehension of high-throughput genomic
data”

• Primarily based on R language (functions can be in other languages),

and run in R environment

• Current release consists of 1741 software packages (sets of

functions) for specific tasks

• Also maintains 948 annotation packages for many commercial arrays

and model organisms plus 371 experiment data packages and 27
workflow packages
Pros of “R” and “Bioconductor”
• FREE

• Open source, not “black box”

• Continual improvements available, cutting- edge statistical methods

• Excellent graphic capabilities

• Available for Windows, Mac OS X and Linux

Cons of “R” and “Bioconductor”
• Difficult to learn

– Mostly command-line interface

– Not as intuitive as point-and-click

– No professional-quality guides and tutorials

• Too many choices sometimes…

• Support not from paid sources, but from community of users

R software versioning
R uses a 3-place versioning system:
3.6.1
Major.minor.revision/patch

Major: only incremented upon substantial code changes or added functionality that could
cause incompatibility with other software designed for previous versions.
Minor: smaller functionality additions and/or substantial bug fixes
Revision/patch: only minor bug fixes

Bioconductor uses the Major.minor 2-place system;

all add-on packages use the 3-place system.
Installing packages in R
Main sources / repositories:
1. Comprehensive R Archive Network (CRAN)
2. Bioconductor

Easiest to do from within R; main methods:

install.packages()
• Gets proper version and any dependency packages
• Default is to only look in CRAN
BiocManager::install() (for R >= 3.5.0)
1. Wrapper around install.packages() to also look in BioC and check for updates to
current packages
2. Must install BiocManager package from CRAN first
3. install_github() in devtools package
 Quiz: If you need Package ‘helloWorld’ version
> 2.43.16, which of below would be OK?

A) 2.8.19
B) 2.45.0
C) 2.43.5
D) More than 1 of the above
E) All of the above
Why do we worry about versions?

• New developments/methods are constantly being added

• Discovered bugs get fixed; no support for problems with
older versions
• If you don’t have R 3.6.1, you’re out of date already!
• Note: if you upgrade to new version of R, you will have to
reinstall your packages from scratch, so make a list first
R development process
• An annual x.y.0 release every Spring* (usually April)

• "Patch" (revision) releases as-needed the rest of the year.

• Each major/minor/revision release is actually a stand-alone installation separate from the

others (but not by default on Macs), with interesting names ("Kite-Eating Tree“, "Short
Summer ", "Another Canoe", "Sincere Pumpkin Patch", "Bug in Your Hair", "Fire Safety",
etc.)

• Most recent release version: 3.6.1 (out 2019-07-05)

• Upcoming patch release: 3.6.2 (schedule 12.02.19)

• Upcoming annual release: 3.7.0 (2020-04-??)

Bioconductor development schedule
• 2 scheduled releases per year: April (a few days after R’s) and
October, usually minor.

• Individual packages in BioC have their own versioning and

patch/revisions are allowed in between.

• Compatibility between BioC packages and R major.minor versions is

crucial; BiocManager::install() function available to automatically get
appropriate package versions!

• Most recent release version: 3.10 (2019-10-30)

 Introduction to R
Studio

The only way you’ll want to use R if

possible
• “Integrated development
environment for R”

• Integrates R console with an

excellent editor, workspace
viewer and graphics
manager

• Make using R *easier*

 EDITOR ENVIRONMENT
HISTORY
“Map” of RStudio

PLOTS
CONSOLE FILE BROWSER
HELP FILES
“Reproducible Research”
• Very important to document how you manipulate / select data,
particularly for large data sets

• R code is an easy way to track what you have done and

instantly reproduce it, even months or years later! (unlike Excel)
• Many tools in the R/Bioconductor community for easy
integration of codes and html output that document both the
codes and the results (ReportingTools, RStudio/Shiny,
RStudio/Rmarkdown/knitr/git)

https://www.rstudio.com/resources/cheatsheets/
 RStudio Text Editor
• Works just like notepad or any
text editor you’re used to

• Creates an exact, reproducible

copy of the commands used

• Text can be saved, edited and

copied

• Comments can be added for

later reference
R Console is where
you execute live
commands
•R has a command-line driven interface;
entered commands or expressions are
evaluated and the proper output is returned
> 2+2
4
> 3*3
9
> log(100)
2
> mean(x)
...
 Key concepts in R coding

1. Objects
1. Hold information in a structured way
2. “class” of the object depends on type and
arrangement of information

2. Functions
1. Pre-written code to perform specific commands
2. Typically used on objects
Common object “classes”
vector – a series of data, all of the same type

matrix – multiple columns of same length, all must have the same type of
data (usually numeric)

data.frame – multiple columns of same length, can be mix of data types,

headers allowed

list – a collection of other objects; each item in the list can be a separate
type of object

function – a command in R
Naming Objects
• In R, use “<-” to create objects; what’s on the left side is the object name and can be
almost anything.

x <- 4

• Object names can consist of letters, numbers, periods* and underscores.

– Cannot start with a number; best to start with letter.
– e.g., x, mydata, mydata_normalized, TrtRep2

• Check to make sure desired object name is not already a function

?objectname

*best practice is to not use ‘.’ because it means something very different in Python
Object attributes

• Standard attributes include dim, class and names

• A matrix is actually a type of vector with ‘dim’ attribute set

• A data.frame is actually a type of list with every item having the same length

• Generic functions like plot() can have methods defined for a particular object
class
How to use functions in R
• Functions are indicated by parentheses – ()
sqrt(81)

• “Arguments” are the input to functions within () and are separated by

commas
ls() 0 arguments
rm(myobject) 1 argument
cbind(x1, x1 + 1) 2 arguments

• Most functions have > 1 argument; input can either be listed in order, or
associated by name.
write.table(object, “outputname.txt”, FALSE)
write.table(object, append = FALSE, file = “outputname.txt”)
Getting help to understand a function
• type in ?rownames
• Anatomy of a help page:
– very top: main.function (package)
– Title
– Sections:
❖ Description
❖ Usage: names arguments in order with (usually) default values
❖ Arguments: description and possible input
❖ Details: further information
❖ Value: the output of the function
❖ ... possibly other sections
❖ Note: any other useful information
❖ References: see for more information, what to cite
❖ See Also: related functions
❖ Examples: how can be used
How to use functions in R, cont.
• R add-on packages - sets of functions that do particular things.

• ONCE only per R version: packages need to be installed

• EVERY time you open R: packages need to be loaded again!

library(edgeR)
❖ “Error in library(edgeR) : there is no package called ‘edgeR’ –
package has not been installed yet

• "Error: could not find function "xxxx" " – package has probably not been
loaded from library.
Functions for exploring objects
str() – overall structure of the object

class() – gives the "class" of the object

length() – gives the number of entries in 1D objects (vector, list)

dim(), nrow(), ncol() – gives number of rows/columns for 2D objects (matrix, data.frame)

names() – gives/sets names of list items or data.frame columns

rownames(), colnames() – gives/sets row & column names of a matrix or data.frame

How R syntax works
• R has strict formats for entering commands and referring to objects; commands that are not
properly formatted will not be evaluated.

• () {} and “” must come in pairs; if not done correctly, R will indicate command is not finished, or will
return error

• R is case-sensitive, except for file names on Windows/Mac

Plot != plot but “myfile.txt” == “MyFile.txt”

• Spaces generally do not matter, except within quotes

temp<-c(1,2) == temp <- c ( 1 , 2 )

• To use \ must use \\ , else use /

Types of R variables

Numeric – 1, 2, 426, 4500, etc.

Character – “a”, “B”, “data”, “cell line”, etc.
Factor – reads as character, but treated as categorical
Logical – TRUE or FALSE
T F
1 0
Missing - NA to indicate missing values
 Base R vs. RStudio / tidyverse (more tomorrow)
• Hadley Wickham wrote a series of packages (the tidyverse) that updates the
way R handles large data to be more data-science friendly

• Way of working/thinking is vastly different:

mtcars$pounds <- mt$wt * 1000 mtcars <- mtcars %>% mutate(pounds = wt / 1000)
•

• Bioconductor pre-dates the tidyverse and many packages don’t work well in
the tidyverse (although this is changing: biobroom, Organism.dplyr,
plyranges )

• Tidyverse can always be used to clean up raw data, even in the downstream
packages don’t “play nice” with it
Subsetting objects in base R
[ and $ are the main base R ways to subset:

• use [ ] to subset 1D objects (vector, list)

• use [ , ] to subset 2D objects (matrix, data.frame)

– rows first, then columns

• inside [ ] or [ , ] can be positions, names in quotes or TRUE/FALSE values.

– Can also be used to re-order objects

• $ can pull out a named column from a data.frame or a named item in a list

• a $ must be followed by the name

Subsetting lists
Given list x:

• x[4:6] is items 4-6

• x[5] is item 5

• x[[5]] is the object in slot 5 (could be another list, dataframe

etc…)

• X[-1] is x without the first entry

Workspace vs. working directory

The Workspace is the internal R memory where it stores the objects you create
during a session. The objects can be saved to/loaded from an external file for
a more permanent copy.

The Working Directory is an external directory (folder) where R will look to

import files or export files when told to given a relative file name (e.g.,
“myfile.txt”). Many functions automatically read from/export to the working
directory without you having to specify it.
Saving work
• Matrix/data.frame objects can be written out to individual files using
write.table() and write.csv()
• All of the R objects can be saved to an .RData file using save.image() or
saveRDS()
– These can only be read by R to re-load objects using load() or readRDS()
– If no filename given to save.image(), will be saved as unnamed .RData
file in current working directory - DO NOT USE
– Note: only objects saved, not how they came to be
• To save all commands entered, can use savehistory(), or...

• Commands more important than objects, so strongly recommend using

RStudio or other text editor to save final, correct version of commands!
Saving work and objects, cont.
• Objects already in the workspace can be overwritten without warning!

• Files in the working directory can also be overwritten without warning!

• There is no "undo" in R other than to rerun the code!

– in RStudio, the "undo" only works on the code editor, not the objects in
the R workspace
Reading in data from tables and sheets

If have Excel-type spreadsheet:

1. Use short column names, no spaces or special characters; do not start
with a number
2. No merged cells, fancy formatting or thousand separators in numbers
3. Save as tab-delimited text file (.txt) or comma separated values file
(.csv) for ease of importing:
❖ read.delim()
❖ read.csv()
R tips and tricks
1. You don't have to understand everything that code does in order to modify it - just
be able to recognize the part that does need modification

2. Search for a package or code snippet that does what you want. Do not “re-invent
the wheel” when you are starting out unless you are doing a learning exercise.

– Example: DO NOT write your own data frame structure, your own statistics
package, your own plotting functions

3. R and Bioconductor packages written by others will often have function(s) that
contain the main computations that you want to do.

4. When a line of code contains multiple computations/functions, run each

computation separately to get a better understanding of what it does or why it may
be throwing an error
How to get help
• Help: ?function or help(function)
for example: ?read.table or ??read.table

• Html help:
1) type help.start()
2) Menu: help -> html help

• To see the code of many functions, simply type the function name. For example: apply

• Base R cheat sheet: http://github.com/rstudio/cheatsheets/raw/master/base-r.pdf

• Longer reference card: http://cran.r-project.org/doc/contrib/Baggott-refcard-v2.pdf

• Understanding R code cheat sheet:

http://go.Illinois.edu/introR
How to get help, cont.
R help mailing list

– https://stat.ethz.ch/mailman/listinfo/r-help

• Bioconductor support site

– https://support.bioconductor.org/

– Be sure to read the posting guide before posting!

– http://www.bioconductor.org/help/support/posting-guide/

• Google!

• Stack overflow
Quitting R
• Before you quit, first save your script in the editor and give it a descriptive
name

• Default prompt asking whether you want to save the workspace image

– If pick "Yes", will save objects in workspace as unnamed .RData file and
commands as unnamed .Rhistory in current working directory; DO NOT
GET IN THE HABIT OF USING THIS!

– If pick "No", will lose objects and codes unless you have saved them
elsewhere; despite risk, this is best for reproducible research unless the R
data objects took a VERY long time to compute, then see above.

– If pick "Cancel", return to R

Additional Resources

https://www.nature.com/news/programming-tools-adventures-with-r-1.16609

https://ropensci.org/packages/

https://www.r-bloggers.com/why-you-should-learn-r-first-for-data-science/

https://www.nature.com/articles/nmeth.3252
Base R ‘SWIRL’ lessons

type :

> install.packages(‘swirl’)

> library(‘swirl’)

>swirl()

Work lessons 1 through 6 at your own pace…

We are available to answer questions!