Introduction to Solving Biological Problems Using R - Day 2

5. Report Writing and Wrap-Up

Creating a markdown file from scratch

So far you have been working with the R notebooks supplied with the course, but for your own analyses you’ll want to start from scratch.

File → New File → R Markdown

• Choose ‘Document’ and the default output type (HTML)
• A new tab is created in RStudio

• The header allows you to specify a Page title, author and output type

  ---
  title: "Untitled"
  author: "Mark Dunning"
  date: "22/12/2016"
  output: html_document
  ---

  Text formatting

  See Help → Markdown Quick Reference in RStudio:

• Enclose text in * to format in italics
• Enclose text in ** to format in bold
• *** for bold italics
• ` to format like code
• $ to include equations: \(e =mc^2\)

• > quoted text:

To be or not to be

• See Help -> Markdown Quick Reference for more:
  • Adding images
  • Adding web links
  • Tables

Not quite enough for a reproducible document

• Minimally, you should record what version of R, and the packages you used.
• Use the sessionInfo() function
  • e.g. for the version of R I used to make the slides

sessionInfo()

Defining chunks

• It is not great practice to have one long, continuous R script
• Better to break-up into smaller pieces; ‘chunks’
• You can document each chunk separately
• Easier to catch errors
• The characteristics of each chunk can be modified:
  • You might not want to print the R code for each chunk
  • …or the output

Chunk options

• It’s a good idea to name each chunk
  • Easier to track-down errors
• We can display R code, but not run it
  • eval=FALSE
• We can run R code, but not display it
  • echo=FALSE
  • e.g. setting display options
• Suppress warning messages
  • warning=FALSE

Chunk options: eval

• Sometimes we want to format code for display, but not execute; we want to show the code for how we read our data, but want our report to compile quickly

e.g.

data <- read.delim("path.to.my.file")

Chunk options: echo

• Might want to load some data, silently, from disk
  • e.g. the R object from reading the data in the previous slide
  • or your P.I. wants to see your results, but doesn’t really want to know about the R code that you used

Chunk options: results

• Some code or functions might produce lots of output to the screen that we don’t need
  • results='hide'

Chunk options: message and warning

• Loading an R package will sometimes print messages and / or warnings to the screen
• This is not always helpful in a report:
• Using message=FALSE and warning=FALSE

library(DESeq)

• Could also need suppressPackageStartupMessages

Chunk options: cache

• The argument cache=TRUE will stop certain chunks from being evaluate if their code does not change
• It speeds-up the compilation of the document
  • we don’t want to reload our dataset if we’ve only made a tiny change downstream

Running R code from the main text

• We can add R code to our main text, which gets evaluated
  • make sure we always have the latest figures, p-values etc
• See example gene-expression-analysis.Rmd

End of Course

Wrap-up

• Thanks for your attention
• Practice, practice, practice
  • … & persevere
• Need inspiration? R code is freely-available, so read other people’s code!
  • Read blogs
  • Follow the forums
  • Download datasets to practice with
  • Bookmark some reference guides
  • on twitter @rstudio, @Rbloggers, @RLangTip
  • Attend the follow-on course on data manipulation and graphics
• Please fill in the feedback form for us to improve the course