MetaSheet-Guard

Documentation website | Repository

MetaSheet-Guard performs experimental-design-aware quality control for sequencing analysis sample sheets. It targets the analysis-preparation stage: after FASTQ generation or public metadata collection, but before running workflows such as Nextflow, Snakemake, nf-core/rnaseq, or custom RNA-seq pipelines.

The current package includes CSV/TSV reading, bundled YAML schemas, validation, repair provenance, workflow export, JSON/HTML reports, and a metasheet-guard CLI.

Scope

MetaSheet-Guard is being built to model relationships between biological samples, sequencing runs, lanes, FASTQ files, replicates, conditions, batches, and downstream workflow requirements. The current release is intentionally small and currently supports these first-pass capabilities:

• required columns
• duplicate column names
• schema-defined column aliases
• empty values in required columns
• sample ID and metadata consistency checks
• FASTQ path, extension, gzip, pair, and duplication checks
• sample/run/lane relationship checks
• batch-condition and related design-risk checks
• safe repair with changes.json provenance
• nf-core/rnaseq, Snakemake, canonical CSV, and DESeq2 design exports
• bundled generic-ngs and bulk-rnaseq schemas

Non-goals

MetaSheet-Guard is not an RNA-seq aligner, quantifier, differential expression tool, SRA downloader, nf-core/fetchngs replacement, nf-schema replacement, Illumina BCL Convert or bcl2fastq SampleSheet validator, single-cell object validator, spatial image validator, or generic CSV validation framework.

Installation

pip install -e ".[dev]"

Quickstart

Validate a broken bulk RNA-seq sample sheet and write a JSON report:

metasheet-guard check examples/broken/missing_required_column.csv \
  --schema bulk-rnaseq \
  --json report.json

The command exits with status code 1 when blocking validation errors are found. For the example above, report.json contains a REQUIRED_COLUMN_MISSING issue because the bulk-rnaseq schema requires a condition column.

Validate a minimal valid example:

metasheet-guard check examples/valid/bulk_rnaseq_paired.csv \
  --schema bulk-rnaseq

Repair safe metadata issues and record provenance:

metasheet-guard repair examples/broken/condition_case_mixed.csv \
  --schema bulk-rnaseq \
  --out clean.csv \
  --changes changes.json

Only safe repairs are implemented. Suggested/inference-based repairs are reserved for later milestones and currently fail clearly if requested.

Export a cleaned sheet:

metasheet-guard export examples/valid/bulk_rnaseq_paired.csv \
  --target nf-core-rnaseq \
  --out nfcore_samplesheet.csv

Python API

from metasheet_guard import read_sheet, validate

sheet = read_sheet("examples/broken/missing_required_column.csv")
result = validate(sheet, schema="bulk-rnaseq")

for issue in result.issues:
    print(issue.severity, issue.code, issue.message)

Development

Run tests and linting:

pytest
ruff check .

The project uses src/ packaging, Typer for the command-line interface, PyYAML for schemas, pytest for tests, and Ruff for linting.