Single-Cell RNA-seq Quality Control

This skill performs comprehensive quality control on single-cell RNA-seq data using MAD-based filtering. It calculates standard QC metrics (total counts, genes detected, mitochondrial percentage), identifies outliers using Median Absolute Deviation, and generates publication-ready visualizations. Follows scverse consortium best practices.

Capabilities

calculate_qc_metrics

Calculate standard QC metrics for each cell

Inputs:

• adata

  (AnnData) (required): Annotated data matrix with raw counts

• mito_prefix

  (string) (optional): Prefix for mitochondrial genes [default: MT-]

Outputs:

• adata

  (AnnData): AnnData with QC metrics in .obs

• qc_summary

  (DataFrame): Summary statistics of QC metrics

filter_cells

Filter cells based on QC metrics using MAD-based thresholds

Inputs:

• adata

  (AnnData) (required): AnnData with QC metrics calculated

• min_genes

  (integer) (required): Minimum genes per cell [default: 200]

• max_genes

  (integer) (required): Maximum genes per cell (None for MAD-based) [default: None]

• max_mito_pct

  (float) (required): Maximum mitochondrial percentage [default: 20.0]

• n_mads

  (float) (required): Number of MADs for outlier detection [default: 5.0]

Outputs:

• filtered_adata

  (AnnData): Filtered AnnData object

• filtering_report

  (dict): Report with cells removed per criterion

generate_qc_plots

Generate comprehensive QC visualization plots

Inputs:

• adata

  (AnnData) (required): AnnData with QC metrics

• output_dir

  (string) (required): Directory to save plots [default: ./qc_plots]

Outputs:

• plot_paths

  (list): Paths to generated plot files

System Prompt

You are an expert bioinformatician specializing in single-cell RNA sequencing analysis.
You follow scverse best practices and understand the biology behind QC metrics.

Key principles:
1. Always examine data distributions before setting hard thresholds
2. Use MAD-based filtering when possible for adaptive thresholds
3. Consider tissue/cell type when interpreting mitochondrial content
4. Document all filtering decisions for reproducibility
5. Generate visualizations to support QC decisions

Examples

Example 1

User: Run QC on my PBMC dataset at /data/pbmc_10k.h5ad

Assistant: I'll perform comprehensive QC on your PBMC dataset. Let me:

1. First, calculate QC metrics (total counts, genes detected, mito %)
2. Examine the distributions to set appropriate thresholds
3. Apply MAD-based filtering (5 MADs from median)
4. Generate visualization plots

[Executes QC pipeline...]

Results:

• Initial cells: 10,000
• Cells removed (low genes): 234
• Cells removed (high mito): 156
• Final cells: 9,610 (96.1% retained)

The filtering thresholds used:

• Min genes: 200
• Max genes: 5,847 (median + 5*MAD)
• Max mito %: 15.2% (median + 5*MAD)

Dependencies

Python Packages

scanpy>=1.9.0
anndata>=0.8.0
matplotlib>=3.5.0
seaborn>=0.12.0
pandas>=1.5.0
numpy>=1.23.0

---

Skill ID:

biomedical.genomics.single_cell_qc

Version: 1.0.0 Category: genomics/single_cell License: MIT Generated: 2025-12-28