Cofunctional grouping-based feature gene selection for unsupervised scRNA-seq clustering
Project description
GeneClust: cofunctional grouping-based feature gene selection for unsupervised scRNA-seq clustering
GeneClust is a computational feature selection method for scRNA-seq cell clustering. GeneClust groups genes into clusters from which genes are evaluated and selected with the aim of maximizing relevance, minimizing redundancy and preserving complementarity.
Dependencies
- numpy>=1.21.5
- pandas>=1.4.2
- anndata>=0.8.0
- setuptools>=59.5.0
- loguru>=0.6.0
- sklearn>=0.0
- scikit-learn>=1.1.1
- scanpy>=1.9.1
- scipy>=1.7.3
- leidenalg>=0.8.9
Installation
- PyPI
You can directly install the package from PyPI.
- Github
Also, You can download the package from Github and install it locally:
git clone https://github.com/ToryDeng/scGeneClust.git
cd scGeneClust/
python3 setup.py install --user
Two Versions of GeneClust
Version | Usage Scenarios |
---|---|
GeneClust-ps | 1. Number of cells is small (e.g., several thousand) 2. Cell clustering performance is more important |
GeneClust-fast | 1. Number of cells is large (e.g., over 50,000) 2. Computational efficiency is more important |
Example Code
from scGeneClust.utils import load_PBMC3k
from scGeneClust import scGeneClust
# load the PBMC3k dataset
raw_adata = load_PBMC3k()
# GeneClust-fast
selected_genes = scGeneClust(raw_adata, version='fast', n_gene_clusters=200, random_stat=2022, verbosity=2)
# GeneClust-ps
selected_genes = scGeneClust(raw_adata, version='ps', n_cell_clusters=7, scale=1000, top_percent_relevance=5, random_stat=2022, verbosity=2)
GeneClust expects raw counts. The output is an ndarray of selected features, which can be used in the downstream cell clustering analysis.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
GeneClust-0.0.1.tar.gz
(23.2 kB
view hashes)
Built Distribution
GeneClust-0.0.1-py3-none-any.whl
(26.5 kB
view hashes)
Close
Hashes for GeneClust-0.0.1-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 236afee59c0bd635bae336fe3033085102283d359872d85b3e8b9cc1e7827b5c |
|
MD5 | 89a6a9bb9b727183ae45934d61944c2a |
|
BLAKE2b-256 | bf045e9786d8a914f2131825f91914020bb39cdb3a0c5d9196d70fa88df07f3d |