search

smorabit / hdWGCNA

High dimensional weighted gene co-expression network analysis
https://smorabit.github.io/hdWGCNA/
Other
315 stars 31 forks source link

Should metacell analysis be conducted when the cell numbers of each cell type range from tens to thousands? #243

Closed zmhsc closed 1 month ago

zmhsc commented 1 month ago

I have a dataset containing over 10,000 cells annotated into a dozen cell types, with the cell numbers of each type ranging from tens to thousands. In the hdWGCNA tutorial, it states, “The dataset used for this tutorial has 40,039 cells, ranging from 890 to 8,188 in each biological sample, and here we used k=25.” Based on this, it appears improper to use k=20 when constructing metacells with my data. Therefore, my question is whether metacell analysis should be conducted and what value should k be set to.

I look forward to your response.

smorabit commented 1 month ago

Yes we always recommend using metacells. I am not sure what value k you should use for your dataset. In general if you have more cells to start with you can use a higher value for k. If I were you I would start with k=50 but you might want to try some values higher or lower as well.

smorabit commented 1 month ago

Closing due to inactivity.