re2cluster

Repetitive unsupervised clustering with optimal parameters by optimizing silhouette score. Based on ARBOLpy package.

Reimplementation of ARBOLpy package by K. Kimler with reduced functionality (see also original R implementation ARBOL)

Installation

Install from github, ideally in conda environment

# conda create -n re2cluster python=3.9 ; conda activate re2cluster
pip install git+https://github.com/lucas-diedrich/re2cluster.git

# Test import 
python 
>>> import re2cluster

Run re2cluster

import scanpy as sc 
import re2cluster 

adata = sc.datasets.pbmc3k()
adata, params, markers = re2cluster.re2cluster(adata)
# verbose output 

Resulting anndata.AnnData object will contain new columns in adata.obs (leiden_tier_<i>, i = 0, ..., n), corresponding to the cluster assignments. Additionally, for every node, the clustering parameters and cluster markers will be returned as dataframes. The package also allows to save QC plots for every node of the iterative clustering procedure.

Uninstall

pip uninstall re2cluster

Missing features

• Stop condition
• Criterium that no clustering is the ideal clustering approach
• Logging/verbosity
• Evaluation functions (adjusted rand index, shannon entropy, gini index)
• plotting functionalities

Reference

@article {Zheng.2021,
		  author = {Hengqi Betty Zheng and Benjamin A. Doran and Kyle Kimler and Alison Yu and Victor Tkachev and Veronika Niederlova and Kayla Cribbin and Ryan Fleming and Brandi Bratrude and Kayla Betz and Lorenzo Cagnin and Connor McGuckin and Paula Keskula and Alexandre Albanese and Maria Sacta and Joshua de Sousa Casal and Faith Taliaferro and Madeline Ford and Lusine Ambartsumyan and David L. Suskind and Dale Lee and Gail Deutsch and Xuemei Deng and Lauren V. Collen and Vanessa Mitsialis and Scott B. Snapper and Ghassan Wahbeh and Alex K. Shalek and Jose Ordovas-Montanes and Leslie S. Kean},
		  title = {A Treatment-Na{\"\i}ve Cellular Atlas of Pediatric Crohn{\textquoteright}s Disease Predicts Disease Severity and Therapeutic Response},
		  year = {2021},
		  doi = {10.1101/2021.09.17.21263540},
		  publisher = {Cold Spring Harbor Laboratory Press},
		  eprint = {https://www.medrxiv.org/content/early/2021/09/22/2021.09.17.21263540.full.pdf},
		  journal = {medRxiv}
}