---

name: motif-pruning-and-regulons description: "Prune pySCENIC co-expression modules against cisTarget ranking databases and convert enriched motif tables into regulons." disable-model-invocation: true metadata: disco-role: operating license: GPL 3.0

Motif Pruning and Regulons

Use this sub-skill when the task is to run or explain the pySCENIC ctx step, call prune2df or find_features, inspect enriched motif tables, convert motif evidence with df2regulons, or choose regulon output formats for downstream scoring.

Do not use this sub-skill for upstream GRN/module inference or downstream AUCell scoring; route those tasks to the pySCENIC sub-skills that own those phases.

Route

• For exact API signatures, motif table schema, output suffix behavior, and interval helpers, read API reference.
• For command-line and Python pruning workflows, no-pruning enrichment mode, resource planning, and cluster modes, read workflows.
• For Feather v2 database, annotation TSV, multiprocessing, empty-output, and extension failures, read troubleshooting.
• To verify local pySCENIC motif/regulon I/O without ranking databases or downloads, run regulon_io_smoke.py with python scripts/regulon_io_smoke.py --help first.

Inputs Owned Here

• Co-expression modules as ctxcore.genesig.GeneSignature or Regulon objects, or CLI module files in CSV, TSV, YAML, GMT, or DAT form.
• cisTarget ranking databases already available on disk, normally current Feather v2 files ending in *.genes_vs_motifs.rankings.feather or *.genes_vs_tracks.rankings.feather.
• Motif-to-TF annotation TSV files with motif id, gene name, motif similarity q-value, orthologous identity, and description columns.

Outputs Owned Here

• Enriched motif tables in CSV or TSV form that preserve NES, AUC, annotation, context, target genes, and rank metadata.
• Regulon/signature collections in GMT, DAT, YAML, or JSON form, with JSON treated as an export mapping rather than the normal pySCENIC reload format.
• ctxcore.genesig.Regulon objects produced by df2regulons for downstream AUCell or custom analysis.

Operating Rules

• Prefer current Feather v2 ranking databases and avoid initiating downloads from this sub-skill.
• Keep ranking databases, motif annotations, and expression/module inputs accessible to every local worker or cluster worker before starting ctx.
• Choose output suffix before running: CSV/TSV when motif evidence must remain inspectable, GMT/YAML/DAT when a compact regulon collection is enough.
• Treat loaded motif CSV/TSV files as trusted inputs because pySCENIC reconstructs Python objects from serialized Context and TargetGenes cells.