Installation¶

Requirements¶

Cherimoya requires:

Python ≥ 3.10
PyTorch ≥ 2.6 (with CUDA support recommended)
A CUDA-capable GPU (for the fused Triton kernel)

Warning

Cherimoya relies on Triton for its custom GPU kernels. Due to its low-level nature, Triton requires a hardware-dependent installation that varies based on your GPU architecture and CUDA version. Please ensure your hardware is compatible and that you have the appropriate drivers and PyTorch-compatible CUDA toolkit installed.

Install from PyPI¶

pip install cherimoya

Install from Source¶

git clone https://github.com/jmschrei/cherimoya.git
cd cherimoya
pip install -e .

Install with uv¶

uv is a fast Python package manager that can be used as a drop-in replacement for pip.

uv pip install cherimoya

Install from source with uv:

git clone https://github.com/jmschrei/cherimoya.git
cd cherimoya
uv pip install -e .

Dependencies¶

The following packages are installed automatically:

Package	Purpose
`torch`	Deep learning framework
`triton`	Custom GPU kernel compilation
`numpy`, `scipy`, `pandas`	Numerical computing and data handling
`h5py`	HDF5 file I/O
`tangermeme`	Genomic sequence utilities and attribution
`bpnet-lite`	Loss functions and logging from BPNet
`macs3`	Peak calling
`bam2bw`	BAM/SAM to bigWig conversion

Optional Dependencies¶

Package	Purpose
`scikit-learn`	Required only for AUROC/AUPRC evaluation metrics
`modisco`	TF-MoDISco motif discovery
`seaborn`	Visualization

Verifying the Installation¶

import cherimoya
print(cherimoya.__version__)

from cherimoya import Cherimoya
model = Cherimoya(n_filters=64, n_layers=9)
print(f"Parameters: {sum(p.numel() for p in model.parameters()):,}")