search

OSCA-source / OSCA.advanced

Advanced steps of the OSCA book
12 stars 10 forks source link

Use same cutoff as docs #5

Closed alexvpickering closed 2 years ago

alexvpickering commented 2 years ago

Wondering if it makes sense to have the same FDR cutoff here as in ?emptyDrops example? Not sure which one you recommend in practice.

LTLA commented 2 years ago

Hm. The example's threshold is probably the one that is less common. A FDR of 1% means that, on average, up to 1% of your filtered cells are empty droplets, which is a bit high when you think about it. I set it to 0.1% to keep the empty droplet frequency so low that I can avoid having to really think about them. Well, in theory, at least.

I'm loathe to change the threshold in the book, as this will change all the downstream results, which in turn requires me to go through the text and update the cluster numbers, etc. You might consider making a PR to update the example instead.

alexvpickering commented 2 years ago

Will do - I mainly raised it to figure out what you recommend. Thank you for the clarification