feat: add reuse_outputs_from config to skip upstream re-computation

When set, creates parse-time symlinks for reference-independent outputs
(pod5/, demux/, bam/rebasecall/, fq/) so only alignment and downstream
rules run. Enables running the pipeline with different references on
identical raw data without re-doing basecalling/demux.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: jayhesselberth

config/config-base.yml

@@ -46,6 +46,11 @@ reference:
   mode: "validate"
   # For build mode only: path to raw tRNA FASTA (without adapters, must end in CCA)
   raw_fasta: null
+  # Optional spacer inserted between CCA and 3' adapter in the reference.
+  # Reads will align with a deletion at this position, absorbing signal
+  # perturbation from the aminoacyl group on charged tRNAs.
+  # Default "" = no spacer (backward compatible). Set to e.g. "T" to enable.
+  cca_spacer: ""
 
 # Optional: provide a kmer table to extract signal metrics using remora
 # Set to a path (e.g. "resources/kmers/9mer_levels_v1.txt") to enable
@@ -123,6 +128,12 @@ report:
   # path to a .qmd file to include at the end of the QC report (optional)
   custom_include: ""
 
+# Reuse reference-independent outputs from a previous pipeline run.
+# Set to the output_directory of a completed run to skip POD5 merging,
+# basecalling, FASTQ extraction, and demultiplexing.
+# Both runs must use the same samples/raw data — only reference can differ.
+reuse_outputs_from: null
+
 warpdemux:
   enabled: false
   barcode_kit: "WDX4_tRNA_rna004_v1_0"

workflow/rules/common.smk

@@ -207,6 +207,32 @@ def find_raw_inputs(sample_dict):
 # set up global samples dictionary to be used throughout pipeline
 outdir = config["output_directory"]
 
+_reuse_from = config.get("reuse_outputs_from")
+if _reuse_from:
+    _reuse_from = os.path.realpath(_reuse_from)
+    # Guard: source must exist
+    if not os.path.isdir(_reuse_from):
+        sys.exit(f"reuse_outputs_from: directory not found: {_reuse_from}")
+    # Guard: must not be same as outdir
+    os.makedirs(outdir, exist_ok=True)
+    if os.path.realpath(_reuse_from) == os.path.realpath(outdir):
+        sys.exit("reuse_outputs_from cannot be the same as output_directory")
+
+    _REUSE_DIRS = ["pod5", "demux", "bam/rebasecall", "fq"]
+    for _subdir in _REUSE_DIRS:
+        _src = os.path.join(_reuse_from, _subdir)
+        _dst = os.path.join(os.path.realpath(outdir), _subdir)
+        if not os.path.isdir(_src):
+            continue  # skip missing (e.g., demux/ when demux disabled)
+        os.makedirs(os.path.dirname(_dst), exist_ok=True)
+        if os.path.exists(_dst):
+            if os.path.islink(_dst) and os.path.realpath(_dst) == os.path.realpath(_src):
+                continue  # already linked correctly
+            sys.exit(f"reuse_outputs_from: {_dst} already exists. Remove it first.")
+        os.symlink(os.path.realpath(_src), _dst)
+        print(f"reuse_outputs_from: {_subdir}/ -> {os.path.realpath(_src)}")
+    del _reuse_from, _REUSE_DIRS
+
 samples = parse_samples(config["samples"])
 samples = find_raw_inputs(samples)