You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have been experimenting using Milo to analyse mass cytometry data. The data I try to feed into the Milo pipeline has been clustered using FlowSOM, and I am using UMAP for 2D projection. My SCE file contains therefore UMAP coordinates (for 10 million cells), but no PCA coordinates. I have tried running the pipeline using the UMAP coordinates when reduced dims are needed, and it works fine until I try to run testNhoods (with glmm); it ends up running for hours without any output (even with only 10 000 cells).
I have also tried to calculate PCA coordinates (without recalculating the UMAP coordinates), and running the pipeline. Now I get some results, however, with adjusted p-values equal to 0.99. Would fewer neighbourhoods (larger k and smaller p) affect the adjusted p-values?
Thank you!
The text was updated successfully, but these errors were encountered:
Hi @NicoFuenza - I would strongly discourage using the UMAP co-ordinates to build a graph as they can distort the distances between individual cells. Could you post some example code and outputs/plots to get a better idea of what you are dealing with concretely.
Hi!
I have been experimenting using Milo to analyse mass cytometry data. The data I try to feed into the Milo pipeline has been clustered using FlowSOM, and I am using UMAP for 2D projection. My SCE file contains therefore UMAP coordinates (for 10 million cells), but no PCA coordinates. I have tried running the pipeline using the UMAP coordinates when reduced dims are needed, and it works fine until I try to run testNhoods (with glmm); it ends up running for hours without any output (even with only 10 000 cells).
I have also tried to calculate PCA coordinates (without recalculating the UMAP coordinates), and running the pipeline. Now I get some results, however, with adjusted p-values equal to 0.99. Would fewer neighbourhoods (larger k and smaller p) affect the adjusted p-values?
Thank you!
The text was updated successfully, but these errors were encountered: