"cellular_prevalence_std" of some "cluster_id" are missing.

[b-niu]

Dear Professor Roth,

I have noticed a fact that, in some cases, the output of write-results-file will make clusters with no cellular_prevalence_std.

Is this expected?

Best regards

[b-niu]

Another example:

[aroth85]

It is not expected. It is likely a numerical issue, when the posterior mass is all allocated to one point i.e. std=0. If you have an example input you could post that would help.

[b-niu]

Thank you professor. Here is the input and output of sample "R025".
pyclone_vi.zip

Wish this will help.

The command used was:

pyclone-vi fit -i R025.pyclone.clean.tsv -o R025.h5 -c 20 -g 100 -r 25 -d binomial

pyclone-vi write-results-file -i R025.h5 -o R025.CCF.tsv