Welcome & Introduction

RAdelaide 2024

Dr Stevie Pederson

Black Ochre Data Labs
Telethon Kids Institute

July 9, 2024

http://blackochrelabs.au/RAdelaide24

Introduction

Who Am I?

Stephen (Stevie) Pederson (They/Them)

• Adelaide, Kaurna Country, SA
• Bioinformatician, Black Ochre Data Labs, Telethon Kids Institute

• Bioinformatician, Dame Roma Mitchell Cancer Research Laboratories (2020-2022)
• Co-ordinator, UofA Bioinformatics Hub (2014-2020)

• PhD (2008-2018) was a Bayesian Model for Transcript-Level Analysis
  • MCMC Engine written in C & R (No R Studio. No Rcpp)

• Best week ever: NAIDOC week, NB Awareness Week + R coding

Who Am I?

Stephen (Stevie) Pederson (They/Them)

• R User for ~20 years \(\implies\) learnt when R was difficult!
• Senior Author of 7 Bioconductor Packages
  • ngsReports, extraChIPs, motifTestR, transmogR
  • strandCheckR, sSNAPPY, tadar

Made countless typos, horrible decisions and catastrophic errors

I crash R at least once a week…

Today’s Tutors

• Dr Jimmy Breen, Dr Liza Kretzschmar & Dr Alastair Ludington (Black Ochre Data Labs)
• Dr Paul Wang, Dr John Salamon (SAGC)
• Dr Na (Charlotte) Sai (University of Adelaide)

Housekeeping

• Toilets are back near the lifts
• Catering will be downstairs in the foyer

Thanks to everyone for sending your information through regarding dietary needs and existing knowledge

Homepage and Material

• The workshop homepage is http://blackochrelabs.au/RAdelaide24
  • Data and course material available here
  • Will stay live in perpetuity
• Links to notes available
  • Slides are directly re-formatted as a simple webpage
  • Slides are visible by clicking the RevealJS link below the TOC
• Group communication can be done through https://bioinformaticshubsa.slack.com/
  • Join the #radelaide24 channel

Course Aims

• Provide a deep understanding of how to work with data in R
  • Importing Data
  • Visualising Data
  • Understanding Data
• Enable use of modern analytic approaches
  \(\implies\) reproducible research
• Not just how \(\implies\) a deep understanding of underlying structures
• The more code you type the more you learn

A Brief Introduction to R

Why use R?

• Heavily used for analysis of biological data (along with Python)
  • Can handle extremely large datasets
    
  • Packages explicitly designed for complex analysis
  • Huge user base of biological researchers
  • (Can be) very fast
• Very easy to dynamically interact with large datasets
  • Can also run as static scripts on HPC clusters

I regularly work with data containing millions of lines

Why use R?

• Reproducible Research!!!
  • Keep records (i.e. scripts) of every step of every analysis
  • Transparent methods
  • Integration with version control such as git
• Avoids common Excel pitfalls (almost) never modify files on disk!

Experience is the best teacher \(\implies\) please practice your skills

• Discuss column sorting with Simon although I believe it’s improved

What is R?

• Derivative of S (Chambers 1977)

What is R?

• Derivative of S (Chambers 1977)
• R first appeared in 1993
  • Ross Ihaka and Robert Gentleman (U of Auckland)
  • Disentangled some proprietary S code \(\implies\) open-source
  • S ceased development in early 2000s (Chambers retired in 2005)
  • Now estimated >2 million users
  • Nice history article here (Chambers 2020)

• Ross Ihaka is of NZ Maori descent
• Last commercial release of S-Plus in 2007

What is R?

• Open source language
  • No corporate ownership \(\implies\) free software
  • Code is managed by the community of users
• R is formally run by a volunteer committee (R Core)
  • Mostly academics
  • John Chambers is still a member
• Annual release schedule + patches
  • Most recent is R 4.4.1 (Jun 14)

Extending R, Chambers (2016)

• Being open source creates headaches for University & Business IT departments
• No guarantees of being virus free. Has inherent security flaws particularly using R data files (which are often part of packages)
• The community self-regulates
• Release Names are references to Peanuts cartoons

R Packages

• Packages are the key to R’s flexibility and power
  • Collections (or libraries) of related functions
  • ggplot2 \(\implies\) Generating plots
  • edgeR \(\implies\) Differential Gene Expression (DGE) for RNA-Seq

R Packages

• \(>\) 16,000 packages are stored on CRAN (https://cran.r-project.org)
  • Not curated for statistical quality or documentation
  • Automated testing for successful installation

• Bioconductor is a secondary repository (https://www.bioconductor.org)
  • \(>\) 2,200 packages with a more biological/genomics focus
  • Curated for language consistency & documentation

• The gg in ggplot2 stands for “Grammar of Graphics”
• Crap packages are generally identified by the users and then just not-used
• Statistical rigour is usually checked during review of the accompanying publication

Helpful Resources

https://r4ds.had.co.nz/

https://r-graphics.org/

Much of today is inspired by a two-day developers workshop I attended with Hadley Wickham. Also gave me an opportunity to have some great conversations with Winston Chang

Using R

The R Console

• Let’s try using R as a standalone tool \(\implies\) open R NOT RStudio
  • On linux: Open a terminal then enter R
  • On OSX: Click on your dock
  • On Windows: Click in your Start Menu
• Do not open

• This is the ‘ugly, old-school’ way of using R
• Still very useful, e.g. testing code on a HPC

The R Console

• This is often referred to as the R Console
• At it’s simplest R is just a calculator (Press Enter)

1 + 1

[1] 2

2 * 2

[1] 4

2 ^ 3

[1] 8

• R has many standard functions

sqrt(2)

[1] 1.414214

log10(1000)

[1] 3

• We place the value inside the brackets after the function name

I never use a calculator program on my laptop, always R

The R Console

We can create objects with names

x <- 5

• We have just created an object called x
• The <- symbol is like an arrow i.e. “put the value 5 into x”
  • Was a single key on keyboards in the 1970s

An APL Keyboard from the 1970s

Object names can be anything but should start with a letter not a number or special character

The R Console

• View the contents of the object by entering it’s name in the Console

x

[1] 5

• The object x only exists in the R Environment

• We can pass objects to functions and perform operations on them

x + 1

[1] 6

sqrt(x)

[1] 2.236068

x^2

[1] 25

x > 1

[1] TRUE

The R Console

• Everything we’ve just done is trivial
• Real analysis isn’t
• If we perform a series of steps
  • Should we keep a copy of what we’ve done?
  • If so, how should we do that?

• A common strategy is to record our code as an R Script
• R Studio makes that easy & convenient
• Many people now use RMarkdown to combine analysis, results and figures

R Studio

Introduction to RStudio

R and RStudio are two separate but connected things

• R is like the engine of your car

• RStudio is the ‘cabin’ we use to control the engine
  • Comes with extra features un-related to R that improve our ‘journey’
  • Known as an IDE (Integrated Development Environment)

• R does all the calculations, manages the data, generates plots
  • i.e. gets us to our destination

• RStudio helps manage our code, display the plots etc
  • i.e. makes our journey easier to navigate

What is RStudio

• RStudio is product of a for profit company (Posit)
  • RStudio (Desktop) is free
  • RStudio Server has annual licence fee of $’000s
• Posit employs many of the best & brightest package developers
  • e.g. tidyverse, bookdown, reticulate, roxygen2 etc.
  • The CEO (JJ Allaire) is still an active developer
• Other IDEs also exist (e.g. emacs, VSCode)

• I remember being at the launch of RStudio (Coventry, 2011). It was a room full of R programmers thinking “holy crap, this changes everything”
• RStudio/Posit is a corporation whilst R is an academic-led volunteer community. So far relatively good relationship
• Heard JJ Allaire present some of his latest work a few weeks ago

Some very helpful features of RStudio

• We can write scripts and execute code interactively
• Predictive auto-completion
• We can see everything we need (directories, plots, code, history etc.)

• Use R Projects to manage each analysis
• Integration with other languages
  • markdown, \(\LaTeX\), bash, python, C++, git etc.
• Numerous add-ons to simplify larger tasks

Important Setup

1. Create a directory on your computer for today’s material
   • We recommend RAdelaide24 in your home directory

2. Now open RStudio
   • RStudio will always open in a directory somewhere
   • Look in the Files pane (bottom-right) to see where it’s looking
   • This is also the working directory for R

We want RStudio to be looking in our new directory (RAdelaide24)
\(\implies\)R Projects make this easy

Create an R Project

(Not needed for any using the Posit cloud)

File > New Project > Existing Directory

• Browse to your RAdelaide24 directory \(\implies\) Create Project

Create an R Project

• The R Project name is always the directory name
• Not essential, but good practice and extremely useful
• The Project Menu is in the top-right of RStudio

• R Projects are simply a wrapper for keeping an analysis organised
  • Will always open in the R Project directory
  • You can easily navigate to a directory with all scripts and data
  • Makes managing file paths from your code very simple
• R Projects can be particularly helpful when loading external files
• Also when saving/exporting lots of files as part of your analysis

Create An Empty R Script

1. File > New File > R Script
2. Save As DataImport.R

RStudio

This is the basic layout we often work with

The Script Window

• This is just a text editor.
• We enter our commands here but they are not executed
• Forms a record of everything we’ve done
  • Can repeat our analysis exactly
• We’ll return here later \(\implies\) but first a quick tour

The R Console

The R Console

• This is the R Console within the RStudio IDE
• We’ve already explored this briefly

• In the same grouping we also have Terminal
  • An approximation of a bash terminal (or PowerShell for Windows)

• Background Jobs shows progress when compiling RMarkdown & Quarto
  • Not super relevant

The R Console

As well as performing simple calculations:

• R has what we call an Environment (i.e. a Workspace)
• We can define objects here or import data
  • Similar to a workbook in Excel with multiple worksheets
  • Much more flexible & powerful
  • Objects aren’t forced to be spreadsheets

• When we create a new sheet in Excel, we’re actually creating an object.
• Most commonly, it’s named Sheet1 or something similar
• Has fixed dimensions for memory management

The R Environment

Like we did earlier, in the R Console type:

x <- 5

Where have we created the object x?

• Is it on your hard drive somewhere?
• Is it in a file somewhere?

• We have placed x in our R Environment
• Formally known as your Global Environment

The R Environment

• The R Environment is like your desktop
• We keep all our relevant objects here
  • Multiple objects are usually created during an analysis
  • Can save all the objects in your environment as a single .RData object
  • R can be set to automatically save your environment on exit

The History Tab

• Next to the Environment Tab is the History Tab
• Keeps a record of the last ~200 lines of code
  • Very useful for remembering steps during exploration
  • Best practice is to enter + execute code from the Script Window

• We can generally ignore the Connections and any other tabs
  • A git tab will also appear for those who use git in their project

Accessing Help

?sqrt

• This will take you to the Help pane for the sqrt() function
  • Contents may look confusing at this point but will become clearer

• Many inbuilt functions are organised into a package called base
  • Packages group similar/related functions together
  • base is always installed and loaded with R
• Click on the underlined word Index at the bottom for a list of functions in the base packages
  • Absolutely no need to learn any of these

• May be issues with URL '/help/library/base/html/00Index.html' not found
• The examples in this help page are a bit rubbish…

Additional Sources For Help

• Help pages in R can be hit & miss
  • Some are excellent and informative \(\implies\) some aren’t
• Bioconductor has a support forum for Bioconductor packages
  • All packages have a vignette (again varying quality)

• Google is your friend \(\implies\) maybe ChatGPT?

As a package author, I’m always reading my own help pages. I simply can’t remember everything I’ve written

The Plots Pane

• We’ve already seen the Files pane
• Plots appear in the Plots pane

plot(cars)

Other Panes

• The Packages Pane is a bad idea
  • Can be disabled by popular request (I always do)
  • Temptation to click is strong
  • Very bad for reproducible research!!!

• Viewer Pane is used when compiling HTML documents from RMarkdown

• Every tab can be minimised/maximised using the buttons on the top right
• Window separators can be be moved to resize panes manually

Cheatsheet and Shortcuts

Help > Cheatsheets > RStudio IDE Cheat Sheet

Page 2 has lots of hints:

• Ctrl + 1 places focus on the Script Window
• Ctrl + 2 places focus on the Console
• Ctrl + 3 places focus on the Help Tab

References

Chambers, John M. 1977. Computational Methods for Data Analysis. New York: Wiley.

———. 2020. “S, r, and Data Science.” Proc. ACM Program. Lang. 4 (HOPL): 1–17.