What problem does it solve? Single-cell ATAC-seq data arrives as raw fragment files that must be quality-filtered, dimensionality-reduced, and converted into peak or gene-activity matrices before any clustering or multiomic analysis can begin. This Skill provides the complete SnapATAC2 preprocessing workflow so you can go from a 10x fragments.tsv.gz file to an analysis-ready AnnData object without re-deriving each step. ## Core Features & Use Cases - Fragment import and QC: Import fragments.tsv.gz directly into AnnData, compute TSS enrichment and fragment-size distributions, and filter cells by counts and TSS score. - Peak calling and matrices: Call peaks de novo with MACS2 or quantify against an existing BED peak set, producing a disk-backed peaks.h5ad file. - Embeddings and gene activity: Run spectral embedding (or TF-IDF + LSI with depth-correlation checks), UMAP, Leiden clustering, and gene-activity scoring against hg38 or mm10 annotations. - Use Case: You receive Cell Ranger ARC output from a multiome experiment and need the ATAC modality embedded and clustered so it can be integrated with RNA data in a MuData object for joint analysis. ## Quick Start Use SnapATAC2 to import my atac_fragments.tsv.gz file, run TSS enrichment QC, build a spectral embedding, and call peaks into a peaks.h5ad file.
