Installation¶
This guide covers installing the aa-tRNA-seq pipeline and its dependencies.
Prerequisites¶
- Operating System: Linux (tested on CentOS/RHEL 9, Ubuntu 20.04+)
- Python: 3.10+
- GPU: NVIDIA GPU with CUDA support (required for basecalling and charging classification)
- Storage: ~50GB for tools, models, and test data
Install Pixi¶
The pipeline uses Pixi for environment management.
| Bash | |
|---|---|
1 | |
| Bash | |
|---|---|
1 | |
| PowerShell | |
|---|---|
1 | |
After installation, restart your shell or run:
| Bash | |
|---|---|
1 | |
For additional installation options, see the official Pixi installation guide.
Clone the Repository¶
| Bash | |
|---|---|
1 2 | |
Install Dependencies¶
Install all Python dependencies via Pixi:
| Bash | |
|---|---|
1 | |
This creates a .pixi directory with all required packages including:
- Snakemake 8.0+
- pysam
- pandas
- pod5
- samtools
- bwa
- deeptools
Install External Tools¶
The pipeline requires several external tools. Install them with a single command:
| Bash | |
|---|---|
1 | |
This downloads and installs:
- Dorado - Oxford Nanopore basecaller (version set in
config/config-base.yml) - Dorado model -
rna004_sup@v6.0.0basecalling model - escpod - POD5 handling, charging classification, and LDX demultiplexing
- WarpDemuX - Barcode demultiplexing (optional, for multiplexed samples)
Dorado and models are installed to resources/tools/ and resources/models/. Modkit is managed by pixi (installed via conda).
Download Test Data (Optional)¶
To run the test pipeline, download the test dataset:
| Bash | |
|---|---|
1 | |
This downloads ~1GB of test POD5 files to .tests/.
Verify Installation¶
Verify everything is installed correctly:
| Bash | |
|---|---|
1 2 3 4 5 6 7 8 9 10 11 | |
Directory Structure After Installation¶
| Text Only | |
|---|---|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 | |
Updating¶
To update the pipeline:
| Bash | |
|---|---|
1 2 | |
To update external tools, modify the version in config/config-base.yml and rerun:
| Bash | |
|---|---|
1 | |
Troubleshooting¶
Pixi Installation Issues¶
If Pixi fails to install, ensure you have:
- curl installed
- Write permissions to
~/.pixi - Internet access to download packages
GPU Not Detected¶
If Dorado fails to detect GPU:
- Check CUDA is installed:
nvidia-smi - Verify CUDA_VISIBLE_DEVICES is set correctly
- Ensure GPU drivers are up to date
escpod Refuses the Charging Model¶
| Text Only | |
|---|---|
1 | |
The installed escpod predates the per-base-feature bundle format. The model
and the runtime are pinned together — escpod_version must be >= 0.19.0 (the floor the charging bundle's basecaller block enforces). Run
pixi run setup to install the pinned version, and check that the escpod on
your PATH is the one under resources/tools/escpod/, not an older one from
~/.local/bin.
Next Steps¶
- Quick Start - Run the test pipeline
- First Analysis - Analyze your own data