Concluding remarks
1 Final notes
Please do the post-workshop survey before you leave! It will really help us improve future workshops.
Material for all lectures, including the external ones have been added to the website. Links are in the schedule on the homepage and in the dropdown menus. The website will stay up indefinitely.
Take a look at the Resources page for links to further resources. We’d like to particularly highlight the Claude courses.
Want to use these tools but want to focus on R? See the Positron setup page for instructions on how to set up and use Positron.
- Let’s hear about what you did in the DIY session!
A lot of our practice prompts were quite basic, covering:
- Writing simple scripts
- Troubleshooting simple errors
- Writing documentation files
You can do much more advanced and ambitious things with Claude Code. Here are some ideas to stay in the realm of the fairly mundane, but quite useful!
Troubleshooting complex errors: e.g., it can be stunningly good at following complex paper trails of errors in a complicated workflow or set of scripts.
Doing things that are (well) beyond your ability and/yet easy to verify:
- A really complicated installation.
- Tedious, repetitive tasks that you don’t have the skill to automate yourself.
Writing entire pipelines, including in formal workflow languages like Nextflow
Proof-reading code (and other documents) that does work.
If you have a complicated set of scripts, it’s very easy to get lost and not notice things like inconsistent variable names or other small errors. These AI tools are often quite meticulous in this sense.Doing “optional next step” stuff you never get to:
- Organize a disorganized project folder, create documentation, and put it under version control
- Turn a script into a nice notebook/Quarto doc, dashboard, or Shiny app
- Turning a set of R scripts into an R package
Build a (Quarto) website
- Personal website
- Lab website
Looking for inspiration for non-coding and/or non-research tasks? Here’s one starting point.