Installation
ST-CNVBench separates the Python controller from the per-method runtime environments.
Choose A Runtime Mode
| Runtime | Best for |
|---|---|
conda |
Local use and method debugging |
docker |
Reproducible runs on systems with Docker support |
apptainer |
HPC systems without Docker daemon access |
Conda Or Mamba
If mamba is available, prefer it for faster dependency solving.
bash conda/install_all_envs.sh
Detailed per-method environment setup is in conda/README.md in the main repository.
Docker
Pull only the images for methods you enable.
docker pull hans0410/cnv-benchmark-copykat:1.1.0
Image names are listed in configs/templates/models.template.yaml. More runtime notes are in External Tools And Runtime Notes.
Apptainer
Create .sif files from the same Docker images and set apptainer_sif in models.yaml.
mkdir -p apptainer_sif
apptainer pull apptainer_sif/copykat.sif docker://hans0410/cnv-benchmark-copykat:1.1.0
Install The Controller
Clone the repository, then create the controller environment from the repository root.
git clone https://github.com/YangLabHKUST/ST-CNVBench.git
cd STCNV-Bench
conda create -n benchmark_env python=3.10 -y
conda activate benchmark_env
pip install -e .
st-cnvbench --help
The package requires Python >=3.10. The initial pip install -e . typically takes about 10-20 min, depending on your machine and network.
Install External Tools
Only a subset of wrappers need external source trees:
CalicoSTClonalscope_NoWGSClonalscope_WGSNumbatXclone
From the repository root, run:
mkdir -p external_tools
cd external_tools
wget https://storage.googleapis.com/broad-alkesgroup-public/Eagle/downloads/Eagle_v2.4.1.tar.gz
tar -xzf Eagle_v2.4.1.tar.gz
rm Eagle_v2.4.1.tar.gz
git clone https://github.com/raphael-group/CalicoST.git CalicoST
git clone https://github.com/seasoncloud/Clonalscope.git clonalscope
git clone https://github.com/kharchenkolab/numbat.git numbat
If you keep the default public layout under external_tools/, the public config templates already point to the expected locations.
Detailed path mapping and model-specific notes are in External Tools And Runtime Notes.
Install Reference Data
Small hg38 annotation files are already bundled in git under refs/hg38_genome_info/.
Large population phasing references are required only for allele-aware wrappers:
CalicoSTNumbatXclone
Download the bundle from:
After download, extract it under:
refs/
└── population_phasing/
Detailed file layout and usage notes are in Reference Data.
Demo Bundle
Download the public cSCC demo bundle from:
After extraction, the expected example outputs should appear under demo_runs/cscc_demo/.
See Quickstart Demo And Expected Outputs for the expected layout and demo commands.
After Installation
- To run the public demo, go to Quickstart Demo And Expected Outputs
- To configure your own benchmark run, go to Dataset Preparation, Model Run, and Evaluation