Feature/variant calling

[Jorisvansteenbrugge]

Add RNA-seq variant calling subworkflow

This PR adds a new subworkflow for variant calling on RNA-seq data, following GATK best practices for RNA-seq short variant discovery.

What's included

• New subworkflow BAM_VARIANT_CALLING that takes a deduplicated BAM and produces a filtered VCF
• Pipeline steps:
  • SplitNCigarReads — splits reads spanning splice junctions
  • HaplotypeCaller — variant calling with --dont-use-soft-clipped-bases
  • MergeVcfs — gathers scattered VCFs per sample
  • VariantFiltration — applies GATK-recommended RNA-seq filters (FS, QD, SnpCluster)
• Scatter-gather parallelization: exon intervals are split into N chunks (configurable via --gatk_interval_scatter_count, default 25) so SplitNCigarReads and HaplotypeCaller run in parallel, drastically reducing wall time
• nf-test coverage including a full run test and stub test for CI
• Bugfixes:
  • The publishDir directive for STAR_ALIGN now correctly excludes bam and fastq files.

Notes for reviewers

• BQSR is intentionally omitted for this initial version. The plan is to first validate the pipeline output without recalibration, and add BaseRecalibrator + ApplyBQSR in a follow-up PR once we've assessed whether it meaningfully improves results for our data
• Variant calling is restricted to exon regions derived from the input GTF (via GTF2BED → BedToIntervalList)
• Default filter thresholds follow GATK RNA-seq recommendations and are exposed as params for customization

Testing

nf-test test subworkflows/local/bam_variant_calling/

[ellendejong]

Nice to see this extension to the pipeline. 👍🏼

[ellendejong]

And some feedback regarding GTF2BED.