Output folder empty after running the ecole_call.sh script

[saakshijalali]

Running source preprocess_samples.sh step creates the Read_Depths and processed_Sample files. But the ecole_call.sh script takes very long to run and the output folder is empty after the script ends without any error. I can see csv files being created in the tmp folder. Kindly suggest possible reason for no files in Output folder

[alperylmzz]

Hi,
This seems like an unusual issue. Typically, the ecole_call.py script deletes all files in the tmp/ folder once the CNV calling process has successfully completed. Could you please share the parameters you used when running ECOLE_call.py ?

Sorry for the very late response

[saakshijalali]

<ecole_call.sh>
python ECOLE_call.py --model ecole --input /ECOLE/processed_samples/ --output /ECOLE/ecole_calls_output --cnv merged --batch_size 16 --normalize ecole_stats.txt -c 0

<ecole_stats.txt>
109.22905835358287, 101.90851300487935

After running the ECOLE_call.py
[ 2024-09-04 12:12:09 ] ECOLÉ: Using CPU!
[ 2024-09-04 12:12:09 ] ECOLÉ: Calling for CNV regions...
[ 2024-09-04 12:12:10 ] ECOLÉ: calling sample: X
[ 2024-09-04 14:14:59 ] ECOLÉ: Calling for regions without read depth information..
[ 2024-09-04 14:14:59 ] ECOLÉ: Processing sample: X

This completes without any error.

[alperylmzz]

There is an issue with the parameter settings for ECOLE_call.py in the ecole_call.sh file. The -c argument is equivalent to --cnv and should be set to either 'merged' or 'exonlevel,' but not '0'. Since you've already assigned --cnv as 'merged,' you can remove the -c 0 part.

[saakshijalali]

Thanks...Let me try as suggested, does the batch size have to be equal to number of Samples?

[alperylmzz]

No, it does not have to be.

[saakshijalali]

The output folder gets created as before but still no Output files generated. The tmp folder has the CSV files created which gets deleted at the end of script.

[alperylmzz]

This issue may arise if an invalid --cnv argument is provided or if the processed_samples folder is empty. At the moment, no other causes come to mind. We’ll continue to investigate possible reasons and keep you updated. In the meantime, could you please try running the script with the --cnv exonlevel option. Let us know if the problem persists.

[saakshijalali]

I was able to successfully run using --cnv exonlevel option and save the results in Output folder. But not able to predict the CNV calls for known positive samples. Do I need to create a baseline as I am using the --model ecole-ft-expert to call CNVs using Exome data.