Detected join operation duplicate emission on right channel

[amizeranschi]

Description of the bug

I am getting the error from the title when running joint germline variant calling with sarek v. 3.6.1 and saving the mapped reads as BAM. This seems to work fine with sarek v. 3.5.1.

Here's an example of the full error message:

Detected join operation duplicate emission on right channel -- offending element: key=[patient:S1, sample:SRR9041541, sex:NA, status:0, n_fastq:12, data_type:cram, id:SRR9041541, num_intervals:17]; value=/home/ubuntu/testSarek/work/d3/2637f9792cb56691e318b196200b8b/SRR9041541_III_1-316620.recal.bai

Command used and terminal output

Relevant files

The attached script should help with reproducing this issue. I have used 2 full WGS samples in yeast, as it has a small genome and run times are still quite manageable for these sorts of tests.

I have tested the following 8 scenarios:

1. Sarek 3.5.1, save as CRAM and enable FASTQ splitting (ran fine)
2. Sarek 3.5.1, save as CRAM and disable FASTQ splitting (ran fine)
3. Sarek 3.5.1, save as BAM and enable FASTQ splitting (failed with error from title)
4. Sarek 3.5.1, save as BAM and disable FASTQ splitting (failed with error from title)
5. Sarek 3.6.1, save as CRAM and enable FASTQ splitting (ran fine)
6. Sarek 3.6.1, save as CRAM and disable FASTQ splitting (ran fine)
7. Sarek 3.6.1, save as BAM and enable FASTQ splitting (ran fine)
8. Sarek 3.6.1, save as BAM and disable FASTQ splitting (ran fine)

sarek-joint-VC-sacCer3.sh

System information

Each of the scenarios that ran successfully took between 20 and 25 minutes to run on a server with 64 CPU cores and 256 GB of RAM, using Nextflow v. 25.04.8. The script requires basic tools like bwa, samtools, gatk and tabix in order to prepare the reference genome annotations.