Course debrief notes - Jan 2023

[tavareshugo]

Debrief notes for the course that ran in Jan 2023 (course page) are detailed below.

Including a checklist of action points here for easier monitoring:

• Adjust schedule timings for day 1 - assign 20 minutes for welcome (hopefully this is caught up by streamlining the cellranger session, see next points)
• cellranger dataset (FASTQ files) needs to be double-checked - maybe subsetting to run faster
• Write instructor guidelines for cellranger section:
  • run through a "hello world" shell script
  • consider switching to gedit or nano as text editors (these are covered in our Intro to Unix course)
• Convert cellranger exercise 1 to a live-coded demo (where participants follow along with the trainer). This should also allow more time for participants to get used to the command line.
• Batch correction diagnostics need to be explained in the slides
• Make slides for day 3 recap - see this jamboard
• Update HTML slides for differential expression/abundance - see these GSlides
• Add grouping of neighbourhoods to the Milo section (possibly this is not covered live in the course, but given as a resource)

---

Day 1

Times: might need to extend the time for the first few sessions and streamline the cellrange session.

• Introduction to Single Cell Technologies - Katarzyna Kania

• Preamble: data set and workflow - Adam Reid

• Library structure, cellranger for alignment and cell calling - Adam Reid

  • Action: Dataset had an issue - check if this is fixed
  • Dropping VS Code? Use nano instead? Or gedit?
  • Action We need to have instructor guidelines about how to deliver this session
    • be explicit that we are trying to teach best-practices as well, so people should include their code in a script and then run it
    • demonstrate a "hello world"
  • Maybe remove exercise 1 and this could instead be done as a live-coded demo (with participants following along).

• Loupe browser demo - Roderik Kortlever

• QC and exploratory analysis - Ashley Sawle

Day 2

• Recap - Adam

• Normalisation - Adam Reid

  • worked well to skip sctransform in favour of starting the Feature Selection in the morning.

• Feature selection and dimensionality reduction - Abigail Edwards

  • this is fine

• Batch correction and data set integration - Abigail Edwards

  • might need better explanation about the diagnostic metrics
  • Action: add diagnostic metrics explanation to slides

Day 3

Timings:

• Need 30 minutes for recap

• Need 1h30 for clustering

• Marker genes spill over to the afternoon (1h extra hour after lunch)

• DE starts at 14:30

• Recap - Hugo

  • Action: Need slides for this

• Clustering - Adam Reid

• Identification of cluster marker genes - Ashley Sawle

• Differential Expression and Abundance Analysis - Hugo Tavares

  • Action: Modified the slides - need to update the HTML (I did it in GSlides)
  • Edited and abridged the Milo section
    • Removed/paraphrased the parts that were verbatin from the Milo docs
    • Didn't include some of the DE downstream analysis: grouping of neighbourhoods and signatures for neighbourhoods
    • Action: add back the grouping of neighbourhoods (might not be covered, but could be there in case it's useful)