Does this skill use hard thresholds or adaptive filtering?

It defaults to MAD-based (Median Absolute Deviation) adaptive filtering to better handle biological variation, but also supports user-defined hard thresholds.

What file formats are supported for single-cell QC?

The skill supports .h5ad (AnnData) and .h5 (10X Genomics Cell Ranger output) files, which are standard in single-cell transcriptomics.

What visualizations are produced during the QC process?

The skill generates distribution plots of QC metrics, threshold overlay plots showing outlier detection, and post-filtering quality summaries.

Can I use this for both human and mouse data?

Yes, you can customize the gene patterns for mitochondrial, ribosomal, and hemoglobin genes (e.g., 'MT-' for human or 'mt-' for mouse) via command-line parameters.

Is this compatible with scanpy and scverse workflows?

Yes, the skill is explicitly designed to follow the latest scverse best practices and integrates seamlessly with scanpy and anndata objects.

Single-Cell RNA-seq Quality Control

Name: Single-Cell RNA-seq Quality Control
Author: anthropics

byanthropics

•

147

Data Science & ML

Performs automated quality control and filtering on single-cell RNA-seq data using scverse best practices.

About

This skill streamlines the preprocessing of single-cell RNA-sequencing data by implementing standardized quality control workflows. It automates the detection of low-quality cells based on count depth, gene detection, and mitochondrial or ribosomal content using Median Absolute Deviation (MAD) thresholds. By following the scverse and scanpy ecosystems' best practices, it generates comprehensive visualizations and filtered datasets, ensuring that downstream analysis—such as clustering and cell-type annotation—is performed on high-quality, reliable data.

Key Features

147 GitHub stars
Flexible filtering parameters for species-specific gene patterns (e.g., MT- vs mt-)
Automated MAD-based outlier detection for counts, genes, and mitochondrial percentages
Support for .h5ad (AnnData) and 10X Genomics .h5 file formats
Comprehensive visualization generation including pre- and post-filtering distributions
Modular Python functions for custom filtering logic and subset-specific QC

Use Cases

Filtering low-quality cells from a new scRNA-seq experiment before downstream analysis
Standardizing preprocessing pipelines across multiple datasets using scverse best practices
Generating standard QC reports and visualizations to assess the quality of a single-cell library

About

Key Features

147 GitHub stars
Flexible filtering parameters for species-specific gene patterns (e.g., MT- vs mt-)
Automated MAD-based outlier detection for counts, genes, and mitochondrial percentages
Support for .h5ad (AnnData) and 10X Genomics .h5 file formats
Comprehensive visualization generation including pre- and post-filtering distributions
Modular Python functions for custom filtering logic and subset-specific QC

Use Cases

Filtering low-quality cells from a new scRNA-seq experiment before downstream analysis
Standardizing preprocessing pipelines across multiple datasets using scverse best practices
Generating standard QC reports and visualizations to assess the quality of a single-cell library