LAP Tracker CellProfiler - ValueError dealing with "no objects"


I had a bug in CP4.0.2 using hte lap tracker: “ValueError all the input arrayw dimensions for the concatenation axis must match exactly…”

Digging down I pinpointed the culprit: my Filtering modul, to remove low expressing H2B nuclei (for classification reasons and so as not to have to remove them by hand in the analysis later on).
Basically, after the filtering there were no objects left

I think this might also have caused another bug I got (after clicking continue processing despite the initial ValueError: "Encountered uncrevorable error in ExportToSpreadsheeet during post-processing: ‘str’ object has no attribute ‘decode’
(note small typo in the error message :slight_smile: )

Now an easy workaround would be to ensure tehre is always an object but this includes data in the CSV file I would actuallw like to filter out from the end pipeline.
Is there a possibility to include this option in the LAP tracker (ie not throw an exception when no objects at frame x but “keep calm and continue trackin’”)?

The full error log from teh console was:




Hi Debbi,

Can you send along that pipeline, here or via other channels? Thanks.

the ‘str’ error is unrelated to the first one - I just got it on a modified version of the pipeline where I think that all frames had objects to track (decreased FilterObject stringency).
It occured at the very end of the analysis in post-processing again:

Worker 10: cellprofiler\modules\ RuntimeWarning: invalid value encountered in true_divide
Worker 10: numpy\core\ RuntimeWarning: invalid value encountered in true_divide
Worker 10: 54

logCP4.0.2_run20200910_16h43.txt (605.0 KB)

The ExportToSpreadsheet error we’ve fixed here, we’ll push a new version likely tomorrow.

The TrackObjects error… I’ve filed a bug report here, but can you tell me if this behavior was the same in 3.1.9 or not? I just ask because the relevant part of the code doesn’t really seem to have changed at all in 4.0; we’ll still give it a shot to fix, I’m just trying to help narrow down where the differences might have crept in so that we can troubleshoot more easily.

A mini-movie (say 3 frames- #1 has objects, #2 has none, #3 the objects are back) and pipeline where this happens would be very helpful for debugging purposes!