This repository provides a reproducible, stepwise SCENIC multirun workflow and downstream regulatory analysis for tumour-associated stroma in colorectal cancer.

If you use this repository, please cite the Zenodo archive below, in addition to the associated Cancer Letters paper:

Howie, J. M. (2026). crc-scenic: SCENIC-based regulatory network analysis of tumour-associated stroma in colorectal cancer (v1.0.0). Zenodo. https://doi.org/10.5281/zenodo.18471338

Formatted citations (BibTeX, APA, etc.) are available via the Zenodo record and GitHub’s “Cite this repository” feature.

This repository implements a three-stage workflow:

1. Prepare SCENIC input loom
   (scripts/01_loom_prep/)

   • Filter and QC a TAS Seurat object
   • Generate a SCENIC-compatible loom
   • Write reusable inputs to resources/

2. Run SCENIC (multirun)
   (scripts/02_scenic_multirun/)

   • Launch SCENIC via a Nextflow pipeline
   • Aggregate results across runs
   • Produce stable regulons and activity scores

3. Downstream integration and analysis
   (scripts/03_downstream/)

   • Integrate SCENIC results into Seurat
   • Visualise regulon activity
   • Identify TAS-specific regulons (RSS)
   • Perform pathway enrichment (ORA)

config/ # Nextflow configuration for SCENIC runs resources/ # Required inputs (SCENIC looms, motif databases) outputs/ # Local, unversioned outputs (figures, tables, logs) scripts/ ├─ 01_loom_prep/ ├─ 02_scenic_multirun/ └─ 03_downstream/

This pipeline starts from a pre-filtered, TAS-only Seurat object derived from colorectal cancer scRNA-seq data. Detailed data provenance (raw data source, TAS re-annotation, and SCENIC pre-filtering) is documented in resources/README.md.

The SCENIC multirun stage is executed via a Nextflow pipeline (see scripts/02_scenic_multirun/), based on an updated fork of the archived VSN pipelines, with minimal changes enabling multirun aggregation using modern cisTarget motif resources: https://github.com/HowieJM/vsn-pipelines