Getting Started¶

Introduction¶

The CARLISLE Pipeline is designed for the analysis of CUT&RUN and CUT&Tag datasets, two closely related chromatin profiling methods that provide high-resolution maps of protein-DNA interactions. Both techniques use antibody-targeted micrococcal nuclease (MNase) or Tn5 transposase to selectively cleave or tag DNA fragments bound by proteins of interest. The resulting DNA is then purified and sequenced to identify binding sites across the genome. CUT&RUN, in particular, offers improved signal-to-noise ratio and lower background compared to traditional ChIP-seq, making it highly efficient even with limited input material.

While CUT&Tag uses an engineered transposase to insert sequencing adapters directly into chromatin fragments, CUT&RUN relies on nuclease digestion and separate library preparation. Despite these differences, the downstream computational workflow—including alignment, filtering, deduplication, and peak calling—is largely similar. The CARLISLE pipeline supports both methods, offering flexibility for diverse chromatin profiling experiments and maintaining compatibility with established Biowulf HPC environments.

Inspired by the nf-core/cutandrun pipeline, CARLISLE incorporates modular and reproducible analysis steps that emphasize transparency and scalability. It begins with raw FASTQ files, performing adapter trimming followed by alignment using Bowtie2. Linear deduplication is an important part of the process—especially for CUT&RUN and CUT&Tag data—to remove PCR artifacts and ensure accurate quantification of unique DNA fragments.

Normalization is performed using either user-provided spike-in controls (e.g., E. coli) or library size-based scaling. Peak calling is then executed using MACS2, SEACR, and GoPeaks—with GoPeaks recommended for its robust performance on CUT&RUN data. Following peak calling, CARLISLE annotates results and summarizes them into detailed reports, with optional differential analysis handled via DESeq2. Quality control metrics are generated using FastQC and MultiQC, while enrichment and annotation are supported through HOMER, ROSE, and ChIP-Enrich.

Setup Dependencies¶

Note: All dependencies are currently module-loaded from the Biowulf HPC environment. However, future versions of the CARLISLE pipeline will transition to a fully containerized setup using Singularity/Apptainer or Docker for improved reproducibility and portability.

CARLISLE relies on several dependencies, most of which are pre-installed and auto-loaded on Biowulf via the ccbrpipeliner module.

Login to the Cluster¶

CARLISLE has been tested and validated on NIH Biowulf HPC.

# SSH into the Biowulf head node
ssh -Y $USER@biowulf.nih.gov

Move to your project directory before proceeding.

Load an Interactive Session¶

An interactive session is required before executing any CARLISLE sub-commands, even if the job will later be submitted to the cluster.

# Request an interactive node
sinteractive --time=12:00:00 --mem=8gb --cpus-per-task=4 --pty bash

Load Dependencies¶

Load the CARLISLE environment by activating the ccbrpipeliner module:

module load ccbrpipeliner/8

Check which version is currently active:

carlisle --version
[+] Loading singularity  4.2.2  on cn0001
[+] Loading snakemake  7.32.4
Pipeline Dir: /vf/users/CCBR_Pipeliner/Pipelines/CARLISLE/.v2.7.1
Version: 2.7.1

carlisle --help
[+] Loading singularity  4.2.2  on cn0001
[+] Loading snakemake  7.32.4
Pipeline Dir: /vf/users/CCBR_Pipeliner/Pipelines/CARLISLE/.v2.7.1
/spin1/home/linux/kopardevn/carlisle
  --> run CARLISLE
  Cut And Run anaLysIS pipeLinE

  USAGE:
    bash /vf/users/CCBR_Pipeliner/Pipelines/CARLISLE/.v2.7.1/carlisle -m/--runmode=<RUNMODE> -w/--workdir=<WORKDIR>
  Required Arguments:
  1.  RUNMODE: [Type: String] Valid options:
      *) init : initialize workdir
      *) run : run with slurm
      *) reset : DELETE workdir dir and re-init it
      *) dryrun : dry run snakemake to generate DAG
      *) unlock : unlock workdir if locked by snakemake
      *) runlocal : run without submitting to sbatch
      *) runtest: run on cluster with included hg38 test dataset
  2.  WORKDIR: [Type: String]: Absolute or relative path to the output folder with write permissions.

  Optional Arguments:
      *) -f / --force: Force flag will re-initialize a previously initialized workdir