cnvkit.py batch with custom pon early exit 0

[oronoc1210]

Hello, I'm having an issue with CNVkit where my batch command with a custom .cnn reference is exiting (with an exit code 0) in the middle of the run, not generating a .cns file.

The runs I performed were as follows:

• generated an antitarget file from my .bed file by running cnvkit.py access into antitarget
• obtained the target and antitarget coverage for each of my normal samples using cnvkit.py coverage
• generated the final my_pon.cnn reference file by running cnvkit.py reference with all of my target and antitarget coverage.cnn files across all of my normal samples
• ran cnvkit.py batch on each of my tumor samples with --method hybrid and --reference my_pon.cnn

For each of the runs across all of my tumor samples, the following files/directories were created:

• .cache/
• .config/
• tumor.antitargetcoverage.cnn
• tumor.targetcoverage.cnn
• my_pon.antitarget-tmp.bed
• my_pon.target-tmp.bed

But not the usual .cns and .cnr files.

This was run with CNVkit v0.9.10, and the stdout was as follows;

CNVkit 0.9.10
Wrote ./TCGA_PoN.cnvkit_pon.target-tmp.bed with 244828 regions
Wrote ./TCGA_PoN.cnvkit_pon.antitarget-tmp.bed with 37695 regions
Running 1 samples in 64 processes (that's 64 processes per bam)
Running the CNVkit pipeline on /keep/70470c6dea515b8e92ee6db153d33a99+7342/TCGA-D5-6538_tumor.dedup.bam ...
Processing reads in TCGA-D5-6538_tumor.dedup.bam
Time: 47.812 seconds (1476023 reads/sec, 5121 bins/sec)
Summary: #bins=244828, #reads=70571815, mean=288.2506, min=0.0, max=12176.08
Percent reads in regions: 56.323 (of 125297503 mapped)
Wrote ./TCGA-D5-6538_tumor.dedup.targetcoverage.cnn with 244828 regions
Processing reads in TCGA-D5-6538_tumor.dedup.bam
Time: 26.452 seconds (916319 reads/sec, 1425 bins/sec)
Summary: #bins=37695, #reads=24238201, mean=643.0084, min=0.0, max=168429.89
Percent reads in regions: 19.345 (of 125297503 mapped)
Wrote ./TCGA-D5-6538_tumor.dedup.antitargetcoverage.cnn with 37695 regions
Processing target: TCGA-D5-6538_tumor.dedup

The whole process took 2 minutes, shortly after the last line the process exits with a return code of 0, which tricked my CWL pipeline into thinking the process was successful, but then failed the overall workflow as it couldn't find the expected but nonexistent .cns file.

Any help with this would be greatly appreciated!

[oronoc1210]

I'm happy to go into detail providing more of the pon generation commands, but the batch command was as follows:

cnvkit.py batch TCGA-D5-6538_tumor.dedup.bam --process 64 --method hybrid --reference TCGA_PoN.cnvkit_pon.cnn

[hiteshi-abv]

Hi CNVkit team,
I am experiencing the same issue. Any help is appreciated. Thank you!

[etal]

The "batch" code when run with multiple processes was eating error messages silently. This is fixed in the new v0.9.13, so you'll see the specific error if you update and run it again now.

If the .cnr files weren't created, then most likely something crashed in the "fix" step. It could have something to do with interval slicing, which was also fixed in the latest release.