Points discussed during sprint #60

lgeistlinger · 2024-10-10T14:55:20Z

Attendees: Andrew Ghazi, Jenny Drnevich, Frederick Tan, Javier Carpinteyro-Ponce, Ludwig Geistlinger

General:

harmonize course summary in learner vs instructor view
make sure every session starts with introductory material and increases difficulty
bulk vs single-cell: intro module required, bulk module beneficial (esp SE and DE)
pointer to cellranger / preprocessing pipeline (= what's not covered before starting)
point out different exercises types in instructor view
how do people run code chunks? (@lgatto: how to pull out all the code chunks of one session into an R script?)
@almahmoud: do workshops need to be re-deployed on the galaxy platform when the code is updated on github or does this happen automatically upon pushing to this github repo?

SCE exercises: (1) plotReducedDim (eg of development stage to tell them a bit of the biology of the dataset), (2) explore count data eg via hist (likely better in EDA/QC)

where does randomness play a role eg in empty drop removal? maybe have one sentence here on the implications of setting / not setting a seed (see also pointer to intro session)
filter on both ends of the distribution of QC criteria, ie not only on the lower ends for library size and number expressed features and on the higher ends for mitochondrial proportion
consider a bit more intro to the general droplet-based experimental procedure to single-cell (10x Chromium figure) to understand QC strategies better
Extension challenges: split very long code chunks and add 1,2 sentences of explanation here and there

maybe a note at the beginning of each session what new packages are loaded in addition to packages that have been introduced / loaded in previous sessions
are there any complications from people having objects from previous sessions in their environments when starting with new sessions?
callout: importance of reference and compatibility of reference and query datasets (-> scDiagnostics)