search

hbctraining / scRNA-seq_online

https://hbctraining.github.io/scRNA-seq_online/.
493 stars 175 forks source link

Regressing out cell cycle scores on SCTransform work-flow #73

Closed camelest closed 1 month ago

camelest commented 2 years ago

Hi, thank you so much for the wonderful resources.

I have a question on calculating cell cycle scores for integrated datasets. Following your tutorials, we are basically doing,

Step1. Normalize by NormalizeData() within each dataset Step2. Cellcyclescoring() within each dataset Step3. Normalize by SCTransform() again within each dataset Step4. Start integration: SelectIntegrationFeatures(), PrepSCTIntegration() etc... And if we want to regress out cell cycle scores, that should be done at Step3.

I have read some threads from Seurat github, and for example, in https://github.com/satijalab/seurat/issues/3665, there is also discussions on alternative workflows such as

Step1. Normalize by SCTransform() within each dataset Step2. Cellcyclescoring() within each dataset Step3. Start integration: SelectIntegrationFeatures(), PrepSCTIntegration() etc... Step4. ScaleData() and regress out cell cycle scores if wanted

Do you have ani thoughts on which way is better? And also if we follow the latter workflow, should the ScaleData() be processed on "RNA" assay?

I'm sorry if this is beyond your online courses, but it would be really helpful if you could add some comments on this. Thank you so much for your help.