Fix: keep sarek status column when all samples are normal (status=0)

[Osamaali313]

Problem

generate_samplesheet.py can silently drop the required status column from a generated nf-core/sarek samplesheet whenever every sample is normal.

In _write_samplesheet, output columns are selected by truthiness:

active_columns = [c for c in column_names if any(c in row and row[c] for row in rows)]

The sarek status column is an integer where 0 = normal, 1 = tumor (config/pipelines/sarek.yaml, described as "critical for somatic calling"). Because 0 is falsy, a cohort in which all samples are normal makes any(... and row[c] ...) evaluate to False, so the status column is omitted from the written CSV.

This is easy to hit in practice:

• Germline-only runs, where every sample is legitimately status=0.
• --no-interactive mode: any sample whose name lacks a tumor/normal keyword defaults to status=0 (_process_sarek_samples), so a directory of plainly-named FASTQs yields an all-normal cohort and loses the column entirely.

The bug is masked by validation: validate_samplesheet runs on the in-memory row dicts (which do contain status), not on the written file — so it reports the samplesheet as valid while the emitted CSV is missing a required column.

Reproduction

rows = [
    {"patient": "P1", "sample": "P1_blood", "fastq_1": ".../P1_R1.fastq.gz", "fastq_2": ".../P1_R2.fastq.gz", "status": 0},
    {"patient": "P2", "sample": "P2_blood", "fastq_1": ".../P2_R1.fastq.gz", "fastq_2": ".../P2_R2.fastq.gz", "status": 0},
]
# before: header = patient,sample,fastq_1,fastq_2      <-- status dropped
# after:  header = patient,sample,fastq_1,fastq_2,status

A mixed cohort (a status=1 present) was unaffected, which is why this slipped through.

Fix

Select columns by explicit presence (value is not None and value != "") instead of truthiness, so valid falsy values are preserved while genuinely-empty columns are still dropped.

Verification

• All-normal cohort → status retained ✅ (previously dropped)
• Tumor-only cohort (all status=1) → status retained ✅
• Mixed cohort → unchanged ✅
• Single-end data (fastq_2 all empty) → fastq_2 still correctly dropped ✅

(The repo has no unit-test harness or pytest config, and CI does not run script tests, so no test file is added — the change is a minimal one-function fix.)

[Copilot]

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

Updates samplesheet column filtering so columns with valid falsy values (e.g., 0) are retained rather than being dropped as “empty”.

Changes:

• Replaces truthiness-based filtering with an explicit “has value” check for samplesheet columns.
• Adds an inline helper (_has_value) to define what counts as “present” data.

---

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