search

grisslab / scAnnotatR

Other
15 stars 2 forks source link

All genes from the cell classifier model must be present in the dataset to perform classification #10

Closed saifsikdar closed 10 months ago

saifsikdar commented 1 year ago

Hi,

I am seeing many warning signs that mention that all genes for each cell classifier model must be present in the dataset to perform classification. Is there any way to classify cells using fewer genes? The code I run is:

seurat.obj <- classify_cells(classify_obj = immune, assay = 'RNA', slot = 'data', cell_types = 'all', path_to_models = 'default') Output: Warning: All genes from T cells classifier model must be present in the dataset to perform classification. Classification of T cells skipped. Error in mat[, idx.chunk, drop = FALSE] : subscript out of bounds

I look forward to hearing your suggestion. Thank you!

jgriss commented 1 year ago

Hi @saifsikdar

Sorry! We missed your question.

Is this still relevant for you?

MarineSr commented 1 year ago

Hi @jgriss, i have the same issue regarding the warning about missign genes. Is there a way to bypass it or to modify the scAnnotate objects ? Thanks a lot, Marine

vintagelego commented 1 year ago

hello,

we're having the same issue here. Could it be due to the fact that our genes are read in a different case (IE CD19 versus Cd19)?

jgriss commented 1 year ago

Hi @vintagelego,

If your genes are in different cases, this generally points to a mouse genome?

Our models are only validated for human data. That's why this does not fit.

Kind regards, Johannes