search

KlugerLab / DAseq

MIT License
39 stars 5 forks source link

Using DAseq after integrating multiple datasets #1

Open hummuscience opened 4 years ago

hummuscience commented 4 years ago

I just stumbled onto your DAseq package and I am eager to give it a try. I am integrating multiple datasets across runs and I am wondering how I can use DAseq on these datasets as I did not see an integration step in your examples in the data.

I am using harmony to do the integration (https://portals.broadinstitute.org/harmony/)

Any ideas? Thanks in advance!

JunZhao1990 commented 4 years ago

DAseq will work on data after integration. For example, in the DAseq tutorial, X.melanoma is the PCA embedding of the data after integration, and this would be the input of the function getDAcells(). Harmony works on PCs directly, so I think you should input harmonized_pcs to DAseq.

Feel free to let me know if this is unclear or you have further questions.

Best, Jun

hummuscience commented 4 years ago

Seems to run well, I got the following error when looking for DA-regions:

Warning message:
In .tclust.warn(O, O$ret) :
  The result is artificially constrained due to restr.fact = 50.
JunZhao1990 commented 4 years ago

This is a warning when DAseq calls the clustering function tclust (package info here). restr.fact is a parameter in the function tclust, which can be passed when calling the function getDAregions. Sometimes, changing restr.fact will lead to different DA regions.