v1.2.2

Bug Fixes

• Fix PeriodicWildcardError in scatter mode (#21, #22): Remove . from the sample wildcard constraint ([A-Za-z0-9_.-]+ → [A-Za-z0-9_-]+). The dot allowed scattered filenames like sample.0000-scattered.annotated.vcf.gz to ambiguously match {sample}.annotated.vcf.gz, triggering infinite wildcard recursion during DAG resolution. This was the true root cause of the previously reported OOM failures — concat now completes in ~67 s using ~2 GB RAM.

• Fix empty Ts/Tv values in QC report (#23): Convert snpsift_tstv and annotation_completeness from run: blocks with shell(read=True) to script: directives using subprocess.run(capture_output=True). The old approach only captured stdout, but SnpSift may write results to stderr, producing empty values. The new scripts capture both streams with a stderr fallback, and also gain proper conda environment isolation.

• Clean up orphaned .tbi index files after scatter/gather (#24): Declare the .tbi auto-created by GATK SelectVariants as a named temp() output on scatter_vcf, preventing ~1 GB of orphaned index files per cohort from persisting after concatenation.

v1.2.1

Bug Fixes

• Fix QC rules failing on SLURM (#19, #20): Replace subprocess.run()/subprocess.Popen() with Snakemake's shell(read=True) in snpsift_tstv and annotation_completeness rules. The subprocess calls spawned a /bin/sh that did not inherit the conda-activated PATH, causing SnpSift: command not found and bcftools: command not found errors when running via the SLURM executor.