Proseg cannot output results even after convergence (MERSCOPE)

[Finley-Maple]

Hi @dcjones , I have successfully run proseg on many different datasets (Xenium, CosMx), but the weird thing is that it cannot output results while finishing the likelihood optimizing process.

The version of Proseg I used is 3.0.8.

Here is the output:

"Resource usage summary:

CPU time : 235952.00 sec.
Max Memory : 35099 MB
Average Memory : 29242.56 MB
Total Requested Memory : 102400.00 MB
Delta Memory : 67301.00 MB
Max Swap : -
Max Processes : 4
Max Threads : 70
Run time : 36007 sec.
Turnaround time : 36003 sec.

The output (if any) follows:

Using 64 threads
Z-coordinate span is zero. Setting voxel layers to 1.
Read dataset:
70510670 transcripts
130591 cells
960 genes
1 fovs"

[Finley-Maple]

Additional info:
Where it got stuck: "counts /../results/proseg_merscope_brain_analysis/merscope_brain_proseg_counts.csv
--overwrite /.../data/MERSCOPE_brain/transcripts_merscope.csv.gz
Using 25 threads
Z-coordinate span is zero. Setting voxel layers to 1.
Read dataset:
70510670 transcripts
130591 cells
960 genes
1 fovs
00:00:00 ############################################################ | log-likelihood: -394467746 / 70510670 (99.82%) | non-background: (96.89%) "

[dcjones]

That's peculiar. Does it exit without generating output, or does it just seem to hang indefinitely after finishing the optimization?

[Finley-Maple]

Apologize for the mistake and possible confusion before.

Actually, the problem lies in the generation of the spatial data output probably due to the lack of z-coordinate, but actually I could get 'transcript-metadata', 'cell-metadata' and 'counts'. When I muted the argument to generate spatial data, it worked.

[alonzowolfram]

Hey @dcjones, just bumping this thread since I'm having similar issues, except with a CosMx data set. I've tested it on the same FOV several times, and sometimes it completes and sometimes it doesn't. I'm only outputting the cell metadata and counts. This is the readout from as far as I get:

Using 28 threads
Finished reading input
Read dataset:
   122992 transcripts
      294 cells
     6466 genes
        1 fovs
00:00:00 ############################################################ | log-likelihood: -1353410.9 | assigned: 104381 / 122992 (84.87%) | non-background: (15.45%)

Any possible workarounds? Thanks!

Lon

[dcjones]

That's interesting, when you say it doesn't complete do you mean it just hangs with this progress bar, or does it crash somehow?

[alonzowolfram]

That's interesting, when you say it doesn't complete do you mean it just hangs with this progress bar, or does it crash somehow?

It just hangs, never crashes.