ATAC-seq Clustering Performance Interpretation

Gjør publiserte scATAC-seq klyngemetrikker om til en rangert metode-sammenligning

av HolobiomicsLab · HolobiomicsLab/asb-skill-collections

Bestått ★ 8.4/10

ATAC-seq Clustering Performance Interpretation — Gjør publiserte scATAC-seq klyngemetrikker om til en rangert metode-sammenligning

Hva den gjør

Et metodekort for å trekke ut NMI, ARI and purity scores out of single-cell ATAC-seq benchmark tables, standardising them into a method-by-dataset matrix, and ranking the pipelines with summary statistics and quantified gaps. Utløses når du are deciding which published clustering method or variant to adopt, or whether yours is competitive. Ren guidance, no scripts or dependencies.

Testrapport

Matet it a four-method by five-dataset NMI table with two cells missing and asked which pipeline to adopt. Both runs picked SnapATAC, but the skill-guided answer added what mattered: it kept the NA cells explicit and reported means twice, over available values and over the three datasets every method shared, which is what exposes that the top raw mean (0.783) belongs to a method that skipped the two hardest datasets and leads by only 0.017 on shared ground. It also produced median, rank, per-dataset winners, a spot-check of five cells against the source, and gaps in absolute and relative terms (+0.124 NMI, +22% for kmers+PCA over motifs+PCA) — the unaided answer gave means, a winner and one caveat. Scope is narrow, essentially one published benchmark, and the frontmatter carries a stray provenance line naming a metabolomics collector script.

Testet på: 2026-07-28 · Claude Code 2.x (agent harness)

Installer

git clone --depth 1 https://github.com/HolobiomicsLab/asb-skill-collections.git
mkdir -p ~/.claude/skills
cd asb-skill-collections && cp -r collections/epigenomics/v1/skills/atac-seq-clustering-performance-interpretation ~/.claude/skills/atac-seq-clustering-performance-interpretation

Kommandoer og eksempelprompter

  • /atac-seq-clustering-performance-interpretationGjør publiserte scATAC-seq klyngemetrikker om til en rangert metode-sammenligning

Skills utløses av vanlige forespørsler — ingen kommandoer å huske. Etter installasjonen aktiverer prompter som disse skillen (på engelsk):

  • Which published scATAC-seq clustering method should I adopt for this?
  • Rank these ATAC-seq clustering methods across datasets using NMI and ARI
  • I have NMI and ARI numbers from a scATAC benchmark, which variant wins?