Problem with allelic_kinetics function

[raul cosentino]

Hi Yuchao,

I am having an issue at the step to generate the “allelic.kinetics.obj”. I get the following message:

Bandwidth 1 : % non-neg estimates 0 corr. freq NA corr. size NA
Bandwidth 2 : % non-neg estimates 0 corr. freq NA corr. size NA
Bandwidth 3 : % non-neg estimates 0 corr. freq NA corr. size NA
Bandwidth 4 : % non-neg estimates 0 corr. freq NA corr. size NA
Bandwidth 5 : % non-neg estimates 0 corr. freq NA corr. size NA
Bandwidth 6 : % non-neg estimates 0 corr. freq NA corr. size NA
Bandwidth 7 : % non-neg estimates 0 corr. freq NA corr. size NA
Bandwidth 8 : % non-neg estimates 0 corr. freq NA corr. size NA
Bandwidth 9 : % non-neg estimates 0 corr. freq NA corr. size NA
Bandwidth 10 : % non-neg estimates 0 corr. freq NA corr. size NA
Error in plot.window(...) : need finite 'xlim' values
In addition: Warning messages:
1: In min(konA.temp, konB.temp) :
no non-missing arguments to min; returning Inf
2: In max(konA.temp, konB.temp) :
no non-missing arguments to max; returning -Inf
3: In min(konA.temp, konB.temp) :
no non-missing arguments to min; returning Inf
4: In max(konA.temp, konB.temp) :


I am not sure what is the problem about it. I generated a subset dataset, which reproduces this result, I am sending all the required objects to run the “allelic_kinetics” function attached, so you can have a look. Is it just that there are too many zero values?? I hope you can help with it. Please, let me know if you need more information.

Best regards,

Raúl

[Yuchao Jiang]

Hi,

The error you encountered was due to SCALE not being able to return non-negative estimates of the bursting parameters, no matter which bandwidth was chosen for the histogram repiling step. SCALE adopted a Poisson-Beta distribution and was by default designed for full-transcript scRNA-seq such as smart-seq and smart-seq2. All of your read counts are below 3 with a significant proportion of zeros. This could lead to the errors. If your data is from 10X Genomics, Drop-seq, or combinatorial indexing (based on the cell names I feel it is), SCALE would not perform well.

You can compare the distribution of the allelic read counts stored in data("mouse.blastocyst")

Hope that this helps,

Yuchao

--
You received this message because you are subscribed to the Google Groups "SCALE_scRNAseq" group.
To unsubscribe from this group and stop receiving emails from it, send an email to scale_scrnase...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/scale_scrnaseq/A99355BA-13E1-4597-99A1-C837355D3F6B%40gmail.com.