satijalab / sctransform

R package for modeling single cell UMI expression data using regularized negative binomial regression
GNU General Public License v3.0
203 stars 33 forks source link

Function future_lapply is using the default value 'future.seed = FALSE' #121

Closed adRn-s closed 2 years ago

adRn-s commented 2 years ago

This gives some warnings when using future::plan("multicore", workers = 32) (for example). As a general rule, it is a good practice to test different seed values when fitting models, and this requires the reproducibility of such a stream of numbers.

After reading this blog post, and this manual details; I understand that changing the default behavior of this function on vst.R and differential_expression.R would be a better practice.

saketkc commented 2 years ago

This has been addressed in the develop branch. Can you try with devtools::install_github("satijalab/sctransform", ref="develop")?

adRn-s commented 2 years ago

Thanks for the clarification.

I know it might not be my place to ask for this, but seeing that the latest release is from Dec 16th 2020, I'd like to ask for you to make a new minor release. To me, it is very important to do my analysis with tagged versions, and I'd rather use a proper number than, say, a commit reference.

saketkc commented 2 years ago

The new version of sctransform (0.3.3) is now on CRAN.