bump viral-ngs 3.0.10, remove GATK IndelRealigner deps, update lofreq pipeline

[dpark01]

Summary

• Bump docker images: viral-ngs 3.0.9 → 3.0.10, py3-bio 0.1.4 → 0.1.5 across all WDL task files
• Remove deprecated --skipRealign: GATK IndelRealigner has been removed upstream from viral-ngs, so the --skipRealign flag on read_utils align_and_fix is no longer needed. Removed the flag and associated skip_realign variable from the align_reads task.
• Update lofreq task: Since GATK IndelRealigner is no longer available, the lofreq task now handles indel normalization internally:
  • Added lofreq viterbi for Viterbi-based read realignment around indels
  • Added lofreq indelqual --dindel to insert indel qualities (required for indel calling)
  • Switched from lofreq call to lofreq call-parallel for parallelized variant calling
  • Added --call-indels flag (controlled by new call_indels Boolean input, default true)
  • Scaled disk size with input BAM size instead of flat 200 GB
  • Increased CPU (2 → 4) and memory (3 GB → 7 GB) for parallel calling
• Bump GitHub Actions: checkout v4→v6, setup-python v5→v6, setup-java v4→v5, setup-buildx-action v3→v4 to resolve Node.js 20 deprecation warnings

Test plan

• miniwdl check passes on all workflows
• CI: validate_wdl_miniwdl passes
• CI: validate_wdl_womtool passes
• CI: test_miniwdl passes (exercises both align_reads and lofreq via assemble_refbased)
• CI: test_cromwell passes
• CI: test_docs passes
• Verify no test expected output updates needed (or update if necessary)

[Copilot]

Pull request overview

Updates the pipeline to use newer viral-ngs/py3-bio container images, removes reliance on deprecated GATK IndelRealigner behavior, and refreshes the intrahost LoFreq calling steps accordingly.

Changes:

• Bump viral-ngs from 3.0.9 → 3.0.10 and py3-bio from 0.1.4 → 0.1.5 across WDL tasks and module requirements.
• Remove deprecated --skipRealign wiring from align_reads now that IndelRealigner is gone upstream.
• Update the lofreq WDL task to realign/normalize indels internally and use lofreq call-parallel, with a new call_indels input.

Reviewed changes

Copilot reviewed 16 out of 16 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
requirements-modules.txt	Bumps referenced module image versions (viral-ngs, py3-bio).
pipes/WDL/tasks/tasks_utils.wdl	Updates docker tags for utility tasks to viral-ngs:3.0.10-* and py3-bio:0.1.5.
pipes/WDL/tasks/tasks_terra.wdl	Updates Terra helper task docker tags to viral-ngs:3.0.10-*.
pipes/WDL/tasks/tasks_taxon_filter.wdl	Updates taxon filter task docker tags to viral-ngs:3.0.10-*.
pipes/WDL/tasks/tasks_sarscov2.wdl	Updates SARS-CoV-2 metadata task docker tags to py3-bio:0.1.5.
pipes/WDL/tasks/tasks_reports.wdl	Updates reporting task docker tags to viral-ngs:3.0.10-* / py3-bio:0.1.5.
pipes/WDL/tasks/tasks_read_utils.wdl	Updates read-utils task docker tags to viral-ngs:3.0.10-core.
pipes/WDL/tasks/tasks_nextstrain.wdl	Updates Nextstrain-related task docker tags to viral-ngs:3.0.10-*.
pipes/WDL/tasks/tasks_ncbi.wdl	Updates NCBI-related task docker tags to viral-ngs:3.0.10-* / py3-bio:0.1.5.
pipes/WDL/tasks/tasks_metagenomics.wdl	Updates metagenomics task docker tags to viral-ngs:3.0.10-classify.
pipes/WDL/tasks/tasks_megablast.wdl	Updates megablast task docker tags to viral-ngs:3.0.10-*.
pipes/WDL/tasks/tasks_intrahost.wdl	Updates LoFreq task to include viterbi realignment + indel quality insertion and switch to call-parallel; bumps docker tag.
pipes/WDL/tasks/tasks_interhost.wdl	Updates interhost task docker tags to viral-ngs:3.0.10-*.
pipes/WDL/tasks/tasks_demux.wdl	Updates demux task docker tags to viral-ngs:3.0.10-* / py3-bio:0.1.5.
pipes/WDL/tasks/tasks_assembly.wdl	Removes deprecated --skipRealign plumbing from align_reads; bumps docker tags to viral-ngs:3.0.10-*.
.github/workflows/build.yml	Bumps GitHub Action versions used by CI workflows.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

call_indels can be set to false, but the workflow still runs lofreq indelqual --dindel and the subsequent resort/reindex steps unconditionally. If indel calling is disabled, these steps are likely unnecessary work and extra I/O; consider guarding indelqual + re-sort/re-index behind call_indels (and otherwise call directly on the viterbi output or the original BAM).

Suggested change

	lofreq indelqual --dindel \
	-f reference.fasta \
	-o indelqual.bam \
	realigned.bam
	# re-sort and re-index after viterbi+indelqual
	samtools sort -o sorted.bam indelqual.bam
	if ~{call_indels}; then
	lofreq indelqual --dindel \
	-f reference.fasta \
	-o indelqual.bam \
	realigned.bam
	# re-sort after viterbi+indelqual
	samtools sort -o sorted.bam indelqual.bam
	else
	# when not calling indels, skip indelqual and sort viterbi output directly
	samtools sort -o sorted.bam realigned.bam
	fi
	# index sorted BAM for variant calling

[Copilot]

The task reserves cpu cores (default 4) but samtools sort is invoked without -@ to use multiple threads. Consider passing -@ ~{cpu} (or a capped value) so the sort step benefits from the allocated CPUs and reduces wall time.

Suggested change

	samtools sort -o sorted.bam indelqual.bam
	samtools sort -@ ~{cpu} -o sorted.bam indelqual.bam