Skip to main content

Stars PyPI PyPIDownloads Documentation Status

rapids-singlecell

Background

This repository offers some tools to make analyses of single cell datasets faster by running them on the GPU. The functions are analogous versions of functions that can be found within scanpy from the Theis lab or functions from rapids-single-cell-examples created by the Nvidia RAPIDS team. Most functions are kept close to the original code to ensure compatibility. My aim with this repository was to use the speedup that GPU computing offers and combine it with the ease of use from scanpy.

Installation

Conda

The easiest way to install rapids-singlecell is to use one of the yaml file provided in the conda folder. These yaml files install everything needed to run the example notbooks and get you started.

conda env create -f conda/rsc_rapids_23.04.yml
# or
mamba env create -f conda/rsc_rapids_23.02.yml

PyPI

As of version 0.4.0 rapids-singlecell is now on PyPI.

pip install rapids-singlecell

The default installer doesn't cover RAPIDS nor cupy. Information on how to install RAPIDS & cupy can be found here.

If you want to use RAPIDS new PyPI packages, the whole library with all dependencies can be install with:

pip install 'rapids-singlecell[rapids]' --extra-index-url=https://pypi.nvidia.com

Please note that the RAPIDS PyPI packages are still considered experimental. It is important to ensure that the CUDA environment is set up correctly so that RAPIDS and Cupy can locate the necessary libraries.

To view a full guide how to set up a fully functioned single cell GPU accelerated conda environment visit GPU_SingleCell_Setup

Documentation

Please have a look through the documentation

Citation

If you use this code, please cite: DOI

Please also consider citing: rapids-single-cell-examples and scanpy

In addition to that please cite the methods' original research articles in the scanpy documentation

If you use the accelerated decoupler functions please cite decoupler

Notebooks

To show the capability of these functions, I created two example notebooks evaluating the same workflow running on the CPU and GPU. These notebooks should run in the environment, that is described in Requirements. First, run the data_downloader notebook to create the AnnData object for the analysis. If you run both demo_cpu and demo_gpu you should see a big speedup when running the analyses on the GPU.

Benchmarks

Here are some benchmarks. I ran the notebook on the CPU with as many cores as were available where possible.

Step CPU (Ryzen 5950x, 32 Cores, 64GB RAM) GPU (RTX 3090) CPU (AMD Eypc Rome, 30 Cores, 500GB RAM) GPU (Quadro RTX 6000) GPU (A100 80GB)
whole Notebook 728 s 43 s 917 s 67 s 57 s
Preprocessing 75 s 21 s 40 s 34 s 30 s
Clustering and Visulatization 423 s 18 s 524 s 27 s 21 s
Normalize_total 252 ms > 1ms 425 ms 1 ms 1 ms
Highly Variable Genes 3.2 s 2.6 s 4.1 s 2.7 s 3.7 s
Regress_out 63 s 2 s 24 s 2 s 2 s
Scale 1.3 s 299 ms 2 s 2 s 359 ms
PCA 26 s 1.8 s 23 s 3.6 s 2.6 s
Neighbors 10 s 5 s 16.8 s 8.1 s 6 s
UMAP 30 s 659 ms 66 s 1 s 783 ms
Louvain 16 s 121 ms 20 s 214 ms 201 ms
Leiden 11 s 102 ms 20 s 175 ms 152 ms
TSNE 240 s 1.4 s 319 s 1.8 s 1.4 s
Logistic_Regression 74 s 4 s 45 s 5 s 3.4 s
Diffusion Map 715 ms 259 ms 747 ms 431 ms 826 ms
Force Atlas 2 207 s 236 ms 300 s 298 ms 353 ms

I also observed that the first GPU run in a new enviroment is slower than the runs after that (with a restarted kernel) (RTX 6000).

Metadata

Release files for rapids-singlecell 0.7.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for rapids-singlecell 0.7.0
File Size Uploaded Provenance Actions
rapids_singlecell-0.7.0.tar.gz 1.1 MB Without provenance Details for rapids_singlecell-0.7.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for rapids-singlecell 0.7.0
File Interpreter ABI Platform Actions
rapids_singlecell-0.7.0-py3-none-any.whl Python 3 none any Details for rapids_singlecell-0.7.0-py3-none-any.whl

Total release size: 2.3 MB

Release files / rapids_singlecell-0.7.0.tar.gz

Download URL rapids_singlecell-0.7.0.tar.gz
Size 1.1 MB
Tags Source
SHA-256 checksum
How to use checksums
d90c0fdbaac333830f5f3797c8cd460fadef3547195ffc6846804cce27e96e6c
BLAKE2b-256 checksum
How to use checksums
c494600c5263fa63f5ffa490621f7640c5a7cf4a3c638d8d3be282199cebaee0
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/4.0.2 CPython/3.9.16

Release files / rapids_singlecell-0.7.0-py3-none-any.whl

Download URL rapids_singlecell-0.7.0-py3-none-any.whl
Size 1.1 MB
Tags Python 3
SHA-256 checksum
How to use checksums
10b0c4562a45d096af373592669e61b271527dbbe3ebe913d466669ac4c2c47b
BLAKE2b-256 checksum
How to use checksums
c2cb5206f1cb83c38207b8d6d650c81db07eb49e44b6a89bcfa7449056887ff0
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/4.0.2 CPython/3.9.16

Security for rapids-singlecell 0.7.0

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.

Release history Release notifications | RSS feed

This release

0.7.0 This release

2 release files No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page