single-cell-rna-qc

To perform single-cell RNA-seq QC on an AnnData h5ad file, this Skill uses MAD-based outlier detection and scverse best practices to compute metrics like total counts, detected genes, and mitochondrial percentage, then exports a filtered dataset.

MAD filtering for single-cell RNA-seq data identifies outliers based on the Median Absolute Deviation of total counts, detected genes, and mitochondrial percentage, removing low-quality cells without relying on arbitrary hard thresholds.

Yes, this single-cell RNA-seq QC pipeline accepts 10X Cell Ranger h5 output files alongside standard AnnData h5ad formats, automatically computing QC metrics and generating threshold visualizations for both.

The best way to visualize mitochondrial percentage thresholds during scRNA-seq quality control is by generating before-and-after plots with threshold overlays, which this Skill produces by following scverse and scanpy visualization best practices.

You do not need to manually set hard cutoffs for mitochondrial percentage when filtering scRNA-seq data, as this Skill primarily uses MAD-based outlier detection, though it optionally supports combining MAD filtering with a hard mitochondrial cutoff.

When running scverse best practices on single-cell RNA-seq data, this Skill computes QC metrics including total counts, detected genes, and MT%, ribosomal, and hemoglobin pattern percentages to identify and remove low-quality cells.