[alejandrogzi]

Why

star_delete_fastq_after_alignment = false had no effect. The STAR and ruSTAR subworkflows passed params.star_delete_fastq_after_alignment ?: true, and Groovy's ?: turns false into true, so the second pass always deleted its input reads.

This matters beyond disk. Each read step deletes its input, which is the output of an earlier task. After a failure, -resume finds those outputs gone and reruns everything from trimming on. On HLdesRot15A spleen (9 samples), xORF's BLAST step failed on an orf panic, and the resume reran 19,153 of 19,177 tasks (1 h 37 min). The panic itself is fixed in hillerlab/xorf#30; modules/xorf moves to that release separately.

Changes

• subworkflows/star_alignment, subworkflows/rustar_alignment: pass params.star_delete_fastq_after_alignment as set. The default stays true in nextflow.config.
• subworkflows/metassembly: BQTOOLS_DECODE (STAR with CBQ reads) and BQTOOLS_ENCODE_ALIGN (bqtools_encode_before_alignment) deleted deacon's output unconditionally. They now follow the same flag, so with false the reads that reach the aligner are kept in every format.
• assets/docs/usage.md (4.6): what each step deletes, and the switches -resume needs. Every other deletion already had a working keep switch: deacon_keep_fastp_fastq, beaver_keep_aletsch_gtf, bigtools_keep_bedgraph, wigtools_keep_bigwig and wigtobigwig_keep_wig.
• nextflow.config: version 0.1.19. Changelog.md: 0.1.19 entry.

With the defaults, the same files are deleted as before.

Testing

• nextflow config -flat for test, test-sb, test-tm, test-polish, test-rustar and test-bqc-star.
• -profile test-bqc-star,docker, which covers bqc, STAR and the CBQ decode step. Each run used Nextflow 25.10.4 on a clean export that includes the xorf submodule:

Settings	verify.py golden numbers	Read files left in work/	Then -resume
Defaults	pass	only the 2 encoded input .cbq; the bqc, deacon and decode outputs are deleted, as before	–
deacon_keep_fastp_fastq = true, star_delete_fastq_after_alignment = false	pass	all: input, bqc and deacon .cbq, decoded FASTQ pairs	29 cached; 28 rerun from RENAME_GTF on (beaver deleted its GTFs); the 4 coverage tasks skipped
all six keep switches	pass	all	61 of 61 cached, none rerun

The middle row is why the usage section lists every switch. With bigtools_keep_bedgraph left out, the cached STAR task has lost its bedGraph, and the resumed run skips coverage without an error.