An analysis framework for modeling dynamical single-cell data with neural differential equations, most notably stochastic differential equations allow us to build generative models of single-cell dynamics.
Quickstart
Please see the scDiffEq website for a quickstart notebook: link
Install the development package
Install generally only takes a few seconds.
Using uv (recommended)
git clone https://github.com/scDiffEq/scDiffEq.git; cd ./scDiffEq;
# Install uv if you haven't already: curl -LsSf https://astral.sh/uv/install.sh | sh
uv sync
Using pip
git clone https://github.com/scDiffEq/scDiffEq.git; cd ./scDiffEq;
pip install -e .
Optional dependency groups
| Extra | Contents |
|---|---|
optional |
Lazily-imported integrations: umap-learn, pillow, ipython, psutil, wandb |
docs |
Sphinx and theme packages for building the documentation |
test |
pytest, for running the test suite |
dev |
Jupyter, ipykernel, and pytest for interactive development |
# Using uv
uv sync --extra docs
# Using pip
pip install -e ".[docs]"
Datasets
import scdiffeq as sdq
adata = sdq.datasets.larry()
Datasets are downloaded on first use and cached under
<data_dir>/scdiffeq_data/. They are hosted on Zenodo
(10.5281/zenodo.21947161); downloads
need no authentication and are verified against the record's md5 checksums.
The record redistributes data published by other groups — please cite the original publications, listed on the data page and in the Zenodo record.
Main API
import scdiffeq as sdq
model = sdq.scDiffEq(adata=adata)
model.fit(train_epochs = 1500)
Built on
System requirements
- Developed on linux20.04 and MacOS (with Apple Silicon), using Python3.11.
- Software dependencies are listed in pyproject.toml.
- Tested with NVIDIA GPUs (A100, T4) and Apple Silicon. Most datasets likely only require an NVIDIA Tesla T4 (free in Google Colab).
Reproducibility
- All results described in the manuscript detailing scDiffEq can be reproduced using notebooks in the companion repository: scdiffeq-analyses
Citation
@article{vinyard2025scdiffeq,
title = {Learning cell dynamics with neural differential equations},
author = {Vinyard, Michael E. and Rasmussen, Anders W. and Li, Ruitong
and Klein, Allon M. and Getz, Gad and Pinello, Luca},
journal = {Nature Machine Intelligence},
volume = {7},
number = {12},
pages = {1969--1984},
year = {2025},
doi = {10.1038/s42256-025-01150-3}
}
If you use the packaged datasets, please also cite their original publications (see the data page).
Metadata
Release files for scdiffeq 1.1.5
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| scdiffeq-1.1.5.tar.gz | 4.1 MB | Without provenance | Details for scdiffeq-1.1.5.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| scdiffeq-1.1.5-py3-none-any.whl | Python 3 | none | any | Details for scdiffeq-1.1.5-py3-none-any.whl |
Total release size: 4.3 MB
Release files / scdiffeq-1.1.5.tar.gz
| Download URL | scdiffeq-1.1.5.tar.gz |
|---|---|
| Size | 4.1 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
f0118b46b2127d7f4b1a2e2a0f5f69ef854a83c3da21d3d9f7da1cbec7904618
|
|
BLAKE2b-256 checksum How to use checksums |
466effbb98589acd41e7a4c1ec6f34d3a3272c9479598a028b839a1e98f11dc9
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
uv/0.13.7 {"installer":{"name":"uv","version":"0.13.7","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"26.04","id":"resolute","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}
|
Release files / scdiffeq-1.1.5-py3-none-any.whl
| Download URL | scdiffeq-1.1.5-py3-none-any.whl |
|---|---|
| Size | 211.5 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
78b2b7791f7eaa6d1b28473215914d69bb0836526c21e07c6a62246eba9336fc
|
|
BLAKE2b-256 checksum How to use checksums |
1938d5a492437b56c839e8160784e74da45e61ddea2a27d4d538597e878abc99
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
uv/0.13.7 {"installer":{"name":"uv","version":"0.13.7","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"26.04","id":"resolute","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}
|
Security for scdiffeq 1.1.5
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.