ELBO value is really low, is it ok?

[jessicaliu70]

cellranger:

here is the log file:
cellbender:remove-background: CellRanger v3 format
cellbender:remove-background: Trimming dataset for inference.
cellbender:remove-background: Including 32084 genes that have nonzero counts.
cellbender:remove-background: Prior on counts in empty droplets is 5
cellbender:remove-background: Prior on counts for cells is 2414
cellbender:remove-background: Excluding barcodes with counts below 5
cellbender:remove-background: Using 16137 probable cell barcodes, plus an additional 26137 barcodes, and 41803 empty droplets.
cellbender:remove-background: Largest surely-empty droplet has 106 UMI counts.
cellbender:remove-background: Running inference...

[sjfleming]

Hi @jessicaliu70 , to me, it looks like this run failed to identify the empty droplet plateau correctly. I am saying that based on seeing that the log says "Prior on counts in empty droplets is 5". I would expect to see a value just over 100, based in the UMI curve you show there.

I might suggest trying

cellbender remove-background --cuda \
    --input your_input_file.h5 \
    --output your_output_file.h5 \
    --expected-cells 10000 \
    --total-droplets-included 40000 \
    --low-count-threshold 50

The --low-count-threshold should help tell cellbender identify the empty droplets, by telling it to ignore barcodes with counts < 50.

[sjfleming]

These kinds of things should be a lot better in v0.3.0

Potentially closed by #238