amalgkit rerun

Overview

amalgkit rerun reads sanity_report.json, resolves failed pipeline targets, and reruns selected commands.

It also writes a manifest so the rerun plan can be inspected or archived.

Basic Flow

Run sanity first:

amalgkit sanity --out_dir ./ --all

Preview rerun targets:

amalgkit rerun --out_dir ./ --dry_run yes

Execute the rerun:

amalgkit rerun --out_dir ./

Default Paths

Report:   out_dir/sanity/sanity_report.json
Manifest: out_dir/sanity/rerun_manifest.json

--metadata report is the default. It uses the metadata path recorded in the sanity report and is usually the safest choice.

Target Selection

If --check is omitted, rerun uses the checks recorded in sanity_report.json.

Supported targets:

• getfastq
• index
• quant
• merge
• busco
• finalize
• all

Rerun selected target families:

amalgkit rerun --out_dir ./ --check getfastq,quant

Limit by run:

amalgkit rerun \
    --out_dir ./ \
    --check getfastq,quant \
    --run SRR000001,SRR000002

Limit by species:

amalgkit rerun \
    --out_dir ./ \
    --check merge,busco,finalize \
    --species "Homo sapiens"

Include warning-level issues, such as missing global summary PDFs:

amalgkit rerun --out_dir ./ --include_warnings yes

Manifest Control

Write the default manifest:

amalgkit rerun --out_dir ./ --manifest inferred

Write to an explicit path:

amalgkit rerun --out_dir ./ --manifest ./rerun_manifest.json

Disable manifest output:

amalgkit rerun --out_dir ./ --manifest none

Operational Notes

• --redo yes is the default so broken outputs can be overwritten.
• --download_dir and --download_lock_dir are shared with download-heavy targets.
• Use --dry_run yes before large cluster reruns.
• Use --include_warnings yes when the goal is to regenerate warning-level missing summaries, not only failed core outputs.