Skip to content

Getting Started

Prerequisites

Pixi

3t-seq uses Pixi to install and manage all software dependencies. Without it you would need to manually install and version-lock Snakemake, conda/mamba, STAR, Trimmomatic, featureCounts, DESeq2, samtools, deeptools, and a dozen other tools — and ensure they all agree on Python and R versions. Pixi replaces that entire process with a single command.

Install it once on your machine (Linux / macOS):

curl -fsSL https://pixi.sh/install.sh | bash

Or follow the official instructions.

Git LFS

3t-seq uses Git Large File Storage (LFS) to store the integration test data (chr19-subset FASTQs and reference files). Without git-lfs those files are downloaded as small pointer stubs, not real data. The setup step will then fail, and the test run will silently process empty inputs.

Install it before cloning:

brew install git-lfs

sudo apt install git-lfs

module load git-lfs

Then activate it globally (once per machine):

git lfs install

Installation

  1. Clone the repository:

    git clone https://github.com/boulardlab/3t-seq.git
cd 3t-seq

    Development branch

    For the latest (potentially unstable) features, clone dev instead: 

    git clone -b dev https://github.com/boulardlab/3t-seq.git

  2. Pull the LFS assets and install all dependencies:

    pixi run -e dev setup   # pulls git-lfs test data
pixi install            # installs all software

Your First Run

Run the integration test to verify the installation:

pixi run test local-references laptop

This processes a 4-sample chr19 subset of the GSE130735 mouse lung dataset (WT vs KO, paired-end) and checks that all pipeline stages complete successfully. Expected runtime: ~10–20 minutes on a laptop.

Step-by-step walkthrough

The Tutorial walks through this exact test run in detail — explaining every config option, sample sheet column, and output file.

Apple Silicon & macOS notes

3t-seq supports macOS including Apple Silicon (M1/M2/M3). A few things to be aware of:

  • GNU coreutils shims: macOS ships BSD versions of ln, sed, etc. that differ from the Linux versions bioinformatics tools expect. When you run the pipeline via Pixi a setup script creates shims in .pixi/macos-shims. Install coreutils via Homebrew so the shims have something to point to:
brew install coreutils
  • Singularity / Apptainer: does not run natively on macOS. Container-based execution requires a Linux VM (e.g., Lima or Colima). Most containers are built for linux/amd64; running them on Apple Silicon adds architecture emulation overhead and may cause instability with Java-based tools.