Tneb clustering

This is the repo with code for "Hierarchical clustering with maximum density paths and mixture models" 2025 paper.
To replicate the images please go through the setup and then go to paper_figures_and_tables/scripts and run the bash script bash_script_to_make_data_and_plots.sh. After this you could also go to the notebooks and replicate the rest of the figures there.

Setup

Clone the repo, initialize the environment and install the package locally. The first line of the yml file sets the new environment's name tneb. This can be overwritten through -n.

git clone https://github.com/ecker-lab/tneb_clustering
cd tneb_clustering
conda env create -f environment.yml 
conda activate tneb 

tsnecude installation

If tsne-cuda installation causes problem, it is likely that your cuda drivers have different version. For more details and installation, see here.

Recreating the densired datasets

Since densired relies on modern packages while some baselines necessitate numpy<2, an additional environment is needed for running the code in the densired.ipynb notebook. It can be installed using conda env create -f environment_densired.yml which creates the environment densired. This step is not necessary to reproduce our results as we provide an precompiled .npz version of the densired datasets.