Questions about the workflow of SPOTlight about spotlight HOT 5 CLOSED

Hi @YMSWhuang

Thank you very much for your interest on SPOTlight! My best guess is that one of the cell types doesn't have any marker genes.
What happens when you do:

unique(as.character(scRNA_mouse$cluster_annotated_general)) %in% unique(all.markers$cluster)

It should be all true since SPOTlight needs marker genes for each cell type to initialize the matrix prior to factorization. If this is the issue you can add a row/s to all.markers with genes representative of the missing population with dummy numbers for avg_log2FC.

Let me know if this helps!

from spotlight.

That's great answer! I checked my scRNA_mouse$cluster_annotated_general and "cluster" in my all.markers, and I found something differences in these cell types. Thanks.

from spotlight.

Hi,
Thanks for this very useful tool your design! I'm having the same question.I don't know how to solve it. Thanks a lot.
The error:

res <- SPOTlight(x = sce,
y = spe,
groups =as.character(sce$cell_type_sub),
mgs =cluster_markers_all_1 , #mgs_df,
#hvg = hvg,
#weight_id = "mean.AUC",
group_id = "cluster",
gene_id = "gene")

Error in trainNMF(x = x, y = y, groups = groups, mgs = mgs, n_top = n_top, :
ids %in% names(mgs) are not all TRUE

names(cluster_markers_all_1)
[1] "p_val" "avg_log2FC" "pct.1" "pct.2" "p_val_adj" "cluster" "gene"

unique(sce$cell_type_sub)%in%unique(cluster_markers_all_1$cluster)
[1] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
[19] TRUE TRUE TRUE TRUE TRUE TRUE

from spotlight.

Hi @SuqinYang ,

My best guess is that the weight_id parameter should be set to avg_log2FC since as you have it currently set up SPOTlight is looking for a weight column.

Hope this helps!
Marc

from spotlight.

The problem is solved. Thank you so much!

from spotlight.