Export to Database Error

cellprofiler

#1

I have recently implemented one of my pipelines to tell CP to skip images with less than 20 cells (this is to avoid measuring cells surviving cytotoxic conditions). The pipeline works well but now when I try to export to database my .h5 file I have a “cannot convert float NaN to integer” error message, which I never had before, and export stops. I GUESS this comes from the fact that lines coming from skipped images are now showing NaN for all values but do you have any suggestion for a workaround?

Thanks

Matteo


#2

Hi, @matteo1. Would you mind sharing your pipeline?


#3

Here you go. Mind that, to save time I usually deactivate the export to database and I just save a .h5 file. I do the “export to database” after image analysis using the same conditions but with the Data Export tool

Q-Screen.cpproj (924.4 KB)


#4

Found this post that dates back in 2013 but might help. I am trying to export to database with “none” selected under “export all objects measurements” and so far so good…If I get it right CPA will only provide per image informations though…


#5

Update: data were exported to an SQLite database OK, but having selected None in the objects measurements to export I only have access to partial informations, therefore not very useful…
Any idea on your side?

Thanks in advance

Matteo


#6

Does anyone have an idea about how to solve this problem?


#7

Hi Matteo,

This sounds like a bug- can you make a GitHub issue with any relevant information and a pipeline and image to reproduce this on our end? Thanks.