Scanpy qc starter

Skill ma-compbio-lab/SkillFoundry/skills/transcriptomics/scanpy-qc-starter

A framework for discovering, compiling, and validating reusable skills for scientific agents.

Install

npx -y skills add ma-compbio-lab/SkillFoundry --skill scanpy-qc-starter

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

What its author says it does

Copied from the file, not written here

Use this skill to run a lightweight Scanpy QC pass on a toy count matrix with the repository-managed Scanpy environment.

SKILL.md

2.8 KB, 681 tokens by cl100k_base, as published. Nobody here has run it

Purpose

Provide a reproducible starting point for Scanpy-based QC work with toy matrix inputs, AnnData export, and a small verification report.

When to use

You want to verify Scanpy and AnnData on this repository.
You need a minimal example layout for single-cell QC work.

When not to use

You need GPU-backed modeling or integrated single-cell workflows.

Inputs

Tab-separated matrix with gene names in the first column and sample or cell columns after that

Outputs

Summary statistics about matrix dimensions and zero fraction
QC summary JSON with per-cell metrics
Optional .h5ad artifact

Requirements

slurm/envs/scanpy
Run with slurm/envs/scanpy/bin/python

Procedure

Inspect examples/toy_counts.tsv.
Run python3 skills/transcriptomics/scanpy-qc-starter/scripts/preflight_counts.py --input skills/transcriptomics/scanpy-qc-starter/examples/toy_counts.tsv.
Run slurm/envs/scanpy/bin/python skills/transcriptomics/scanpy-qc-starter/scripts/run_scanpy_qc.py --input skills/transcriptomics/scanpy-qc-starter/examples/toy_counts.tsv --summary-out skills/transcriptomics/scanpy-qc-starter/assets/scanpy_qc_summary.json --h5ad-out scratch/scanpy/toy_counts_qc.h5ad.
Inspect total_counts, n_genes_by_counts, and pct_counts_mt in the summary output.

Validation

Preflight script exits successfully.
Reported dimensions match the example matrix.
Scanpy writes a valid .h5ad file.
QC summary includes per-cell metrics.

Failure modes and fixes

Ragged rows: normalize the input to a rectangular TSV.
Non-numeric counts: clean or coerce the count columns before QC.
Missing Scanpy environment: recreate slurm/envs/scanpy and rerun with that interpreter.

Safety and limits

This is a lightweight QC starter, not a full single-cell analysis workflow.
Toy matrices are for environment verification and examples only.

Examples

python3 .../preflight_counts.py --input skills/transcriptomics/scanpy-qc-starter/examples/toy_counts.tsv
slurm/envs/scanpy/bin/python .../run_scanpy_qc.py --input skills/transcriptomics/scanpy-qc-starter/examples/toy_counts.tsv --summary-out scratch/scanpy/summary.json --h5ad-out scratch/scanpy/toy_counts.h5ad

Provenance

Scanpy docs: https://scanpy.readthedocs.io/en/stable/
Scanpy installation docs: https://scanpy.readthedocs.io/en/stable/installation.html
Scanpy QC metrics API: https://scanpy.readthedocs.io/en/stable/generated/scanpy.pp.calculate_qc_metrics.html
AnnData docs: https://anndata.readthedocs.io/en/latest/index.html
scvi-tools docs: https://docs.scvi-tools.org/en/latest/index.html

Related skills

snakemake-toy-workflow-starter