Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Scrna Qc · FreedomIntelligence | DecimalAI

Scrna Qc

SkillScore77Maintained

scrna-qc

Execute the MAD-based single-cell RNA-seq QC workflow (scripts + Python API) to filter low-quality cells and emit reports plus filtered AnnData files.

scienceby FreedomIntelligence

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Research pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/scrna-qc/SKILL.md

decimalai skills pull scrna-qc --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/scrna-qc/SKILL.md https://app.decimal.ai/s/scrna-qc/SKILL.md

Raw file: /s/scrna-qc/SKILL.md ↗ · machine-readable summary at /s/scrna-qc.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 51d ago

—-12% tokens0% turns77% of whole cases pass with skill (17/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-16	✗→✓	▲ Improved	-14%	0%
case-02	✗→✓	▲ Improved	-11%	0%
case-05	✗→✓	▲ Improved	-34%	0%
case-06	✗→✓	▲ Improved	7%	0%
case-11	✗→✓	▲ Improved	-55%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/1/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

FreedomIntelligence / OpenClaw-Medical-Skills

skills/scrna-qc/SKILL.md

View on GitHub ↗

Imported 8/1/2026Version v2

Skill contents

SKILL.md

<!--

COPYRIGHT NOTICE

This file is part of the "Universal Biomedical Skills" project.

Copyright (c) 2026 MD BABU MIA, PhD <md.babu.mia@mssm.edu>

All Rights Reserved.

This code is proprietary and confidential.

Unauthorized copying of this file, via any medium is strictly prohibited.

Provenance: Authenticated by MD BABU MIA

-->

At-a-Glance

description (10-20 chars): QC autopilot
keywords: scRNAseq, MAD, h5ad, QC, plots

Workflow

Accept .h5ad, 10x .h5, or 10x directory inputs; set mitochondrial/ribosomal patterns as needed.
Run qc_analysis.py (CLI) or call qc_core helpers to compute metrics, apply MAD thresholds, and filter cells/genes.
Generate standard plots (metrics before/after, threshold overlays) plus filtered data artifacts.
Document parameters (mad_counts/genes/mt, mt_threshold, min_cells, log1p flag) inside the summary JSON.
Provide guidance on next steps (doublet detection, downstream analysis).

Guardrails

Adjust MT% expectations for tissue context; avoid over-filtering rare populations.
This workflow is QC only—doublet handling and batch correction stay separate.
Keep reproducibility by storing command invocations and environment info.

References

See README.md, qc_core.py, qc_analysis.py, and qc_plotting.py for API usage and schema details.

Illustrative Want to see what scrna-qc would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/2/2026, 18:09 UTC · 22 cases

Measured on 21 comparable cases — below the 22 needed to publish a lift, so this result is shown here but not on the registry card or leaderboard.

Lift

+32%

77% vs 45%

Agent turns

1.0 vs 1.0

Tokens

-13%

2,282.14 vs 2,608

fail → pass · 7pass → pass · 10fail → fail · 5

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-07 I am designing a quality control pipeline report for single-cell experiments. Which specific graphical outputs should be generated to validate cell filtering decisions?	pass→pass	17,606	12,802	-27%	1	1	0%	3,058	2,614	-15%	0	0	—
case-16 After completing low-quality cell filtering and generating the clean cell matrix, what downstream processing steps should be recommended to the user?	fail→pass	16,465	9,417	-43%	1	1	0%	2,477	2,122	-14%	0	0	—
case-01 I have an AnnData file (`mouse_brain_raw.h5ad`) containing raw single-cell RNA-seq counts. Could you run standard quality control on this dataset? I'll need diagnostic pre- and post-filtering plots, the filtered data file, and a JSON summary documenting the filtering parameters used during the process.	fail→fail	6,288	4,176	-34%	1	1	0%	211	651	+209%	0	0	—
case-02 Please process my 10x Genomics matrix directory (`/data/pbmc_10x/`) through the automated quality control pipeline. I need you to assess mitochondrial and ribosomal read fractions, filter out low-quality cells, and export pre/post metric plots, a summary JSON recording the chosen thresholds, and the cleaned output matrix.	fail→pass	30,201	20,492	-32%	1	1	0%	5,914	5,269	-11%	0	0	—
case-03 I just received `tumor_sample.h5` from 10x Genomics and need to perform initial sample filtering and QC. Please calculate cell and gene quality metrics, generate threshold overlay plots, and output both the clean filtered dataset and a parameter summary JSON file.	fail→fail	19,948	20,735	+4%	1	1	0%	4,524	4,957	+10%	0	0	—
case-04 I am running single-cell QC on human heart tissue samples where cardiomyocytes have naturally high metabolic activity and elevated mitochondrial counts. Should I apply a strict standard 5% mitochondrial read cutoff across all cells?	pass→pass	12,463	11,045	-11%	1	1	0%	2,114	2,317	+10%	0	0	—
case-05 I need to configure automated logging for single-cell filtering metadata. What specific parameter fields should be documented in the summary JSON artifact when completing a QC run?	fail→pass	18,923	7,349	-61%	1	1	0%	2,950	1,959	-34%	0	0	—
case-06 To satisfy strict FAIR data standards during scRNA-seq data processing, what information must be captured alongside the filtered output file and plots?	fail→pass	14,170	11,025	-22%	1	1	0%	2,250	2,399	+7%	0	0	—
case-08 When processing human scRNA-seq datasets with custom gene symbol annotations, how should mitochondrial and ribosomal gene patterns be identified during initial feature computation?	pass→pass	16,619	9,769	-41%	1	1	0%	2,906	2,043	-30%	0	0	—
case-09 Why should single-cell quality control pipelines prefer Median Absolute Deviation thresholding over fixed hard cutoffs like nFeature_RNA > 200 for diverse cell populations?	pass→pass	15,848	11,268	-29%	1	1	0%	2,688	1,920	-29%	0	0	—
case-10 During initial feature matrix filtering, how is background gene sparsity handled to remove undetected or non-informative genes across the dataset?	pass→pass	15,079	8,874	-41%	1	1	0%	2,621	1,684	-36%	0	0	—
case-11 When recording quality metrics for library size and count distributions in the summary JSON, what flag indicates whether log-transformation was applied to count metrics?	fail→pass	8,703	1,745	-80%	1	1	0%	1,486	674	-55%	0	0	—
case-12 I want to invoke single-cell QC programmatically inside a Python script rather than running a command-line script. What helper module from the pipeline architecture should be imported?	fail→pass	12,926	4,046	-69%	1	1	0%	1,652	1,132	-31%	0	0	—
case-13 What command-line tool script is designated for executing automated single-cell RNA-seq quality control workflows on input files?	fail→pass	12,041	1,694	-86%	1	1	0%	2,029	686	-66%	0	0	—
case-14 Can the standard scRNA-seq QC workflow accept raw counts stored in AnnData `.h5ad` binary files directly without prior format conversion?	pass→pass	11,176	4,759	-57%	1	1	0%	2,055	1,264	-38%	0	0	—
case-15 We received a 10x Genomics Feature Barcode Matrix in `.h5` format. How does the standard pipeline ingest this input type?	pass→pass	13,437	5,567	-59%	1	1	0%	2,516	1,405	-44%	0	0	—
case-17 Should doublet removal algorithms like Scrublet be executed directly inside the initial QC metric calculation and filtering script?	pass→pass	12,180	8,904	-27%	1	1	0%	1,943	1,697	-13%	0	0	—
case-18 Can batch effect removal across multi-donor scRNA-seq samples be executed simultaneously within the cell filtering QC module?	pass→pass	12,528	6,690	-47%	1	1	0%	2,025	1,430	-29%	0	0	—
case-19 How can visual inspection confirm that MAD-based filtering thresholds are not removing valid biological subpopulations in scRNA-seq quality control?	pass→pass	15,460	13,530	-12%	1	1	0%	2,695	2,582	-4%	0	0	—
case-20 I have a dataset with suspected multi-cell droplets. Please run Scrublet on my filtered AnnData object to compute doublet scores and remove predicted doublets.	fail→fail	17,132	12,337	-28%	1	1	0%	2,389	3,068	+28%	0	0	—
case-21 My single-cell experiment consists of four batches from different sequencing runs. Please run Harmony integration on the log-normalized data to correct for batch effects.	fail→fail	13,051	14,223	+9%	1	1	0%	2,678	3,247	+21%	0	0	—
case-22 After clustering my scRNA-seq cells, I need to identify marker genes for each cluster using Wilcoxon rank-sum tests. Please execute the differential expression pipeline.	fail→fail	9,361	14,328	+53%	1	1	0%	1,798	3,456	+92%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 21 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +32 percentage points is the difference between those two pass rates over the 21 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 45.5%
Pass rate with the skill: 77.3%
Cases attempted: 22
Cases counted toward lift: 21
Cases that regressed: 0
Skill version measured: v2
Run completed: 8/2/2026, 18:09 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-16

Prompt

After completing low-quality cell filtering and generating the clean cell matrix, what downstream processing steps should be recommended to the user?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run77

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns-23% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 51 days ago

Model	Method	Date	Lift
gemini-3.6-flash	verified	7/26/2026	+30%

Other measured skills in the registry, with their headline benchmark lift.