Input with 10x matrix containing Antibody Capture

[ollieeknight]

Heya,

I'm getting an error that the feature names don't match, from a 10x 5' run where gene expression and cite-seq are counted together. I'm using it through the command line, like:

celltypist --transpose-input --indata outs/filtered_feature_bc_matrix/matrix.mtx.gz --model Immune_All_Low.pkl --outdir celltypist/ -gf outs/filtered_feature_bc_matrix/features.tsv.gz -cf outs/filtered_feature_bc_matrix/barcodes.tsv.gz

I'm wondering if the features don't match because the .mtx file has two entries, Gene Expression and Antibody Capture. Is there a way to subset the input so that only the first matrix (Gene Expression) is chosen?

Thanks!

Ollie

[ChuanXu1]

@ollieeknight, does your features.tsv file contain two columns? It is supposed to only have one feature column (without header).

[ollieeknight]

thanks for getting back to me - it has three columns (ensemble ID, gene name, matrix present): features1.csv. I had to convert it to a csv to upload it here, but it was originally a .tsv.gz file.

[ChuanXu1]

@ollieeknight, you may need to manually delete the first and third columns to get a new table as the feature input file.

[ollieeknight]

no worries, thanks for letting me know. this is just a standard 10x cellranger count output, so might it be worth adding this functionality in the future? Just a thought. Appreciate your help!