Closed auberginekenobi closed 2 weeks ago
Thanks Owen, this is a good find, I will take a look ASAP.
Sure thing, a related note that CNVkit recommends to pool all normal samples to generate the reference but that's maybe a feature request or an improvement rather than strictly related to the bugfix
Delayed response on my part as I have been out-of-office.
Temporary workaround is to keep one of the normals as your "Normal" sample (recommend normal blood if possible) and set the other additional normals to "Tumor" status. This will use the one normal sample as the matched normal for all samples, including the additional normal samples. Not perfect but okay if you expect your multiple normal samples to be fairly similar.
Hi Owen, if you pull the latest version of AmpliconSuite-pipeline from GitHub you can now access the latest version of AmpliconSuite-pipeline that fixes the issue with multiple normal files in GroupedAnalysis (v1.3.4).
Version 1.3.4 released and bug fixed in this version
Repro:
Running AS-p GroupedAnalysis mode via registry.hub.docker.com/jluebeck/prepareaa:v1.3.3:
python /home/programs/AmpliconSuite-pipeline-master/GroupedAnalysisAmpSuite.py --input input_file.txt --output_directory redacted --nthreads 4 --ref hg38 --cngain 4.5 --rscript_path /usr/bin/Rscript --cnvkit_dir /usr/local/bin --python3_path /usr/bin/ --AA_src /home/programs/AmpliconArchitect-master/src > job.stdout.log
Contents of input_file.txt:
Expected behaviour: program terminates successfully, produces AA outputs for each tumor sample.
Actual behaviour: program encounters fatal error while creating cmd strings for AA runs. stdout:
stderr:
Diagnosis:
On line 356 of
GroupedAnalysisAmpSuite.py
, the seed for the first normal sample is set correctly but no other normal sample records are altered. Then, if the--skip_AA_on_normal_bam
flag is not set, AA runs are initiated for all tumor and normal samples. However, the correct argstrings have only been generated for all tumor and the first normal samples, leading to the error.