Disclaimer

This is an unofficial Python implementation for the R package EcoTyper, for discovery of ecotypes in scRNA-seq data, and recovery in bulk RNA-seq. The original code has been optimized for sparse matrices and smoother integration with AnnData. The NMF solver can now be changed to the much quicker Coordinate Descent algorithm.

Installation

This package is not yet available on PyPi. To install ecotyper for Python, clone the repository

$ git clone https://github.com/hoseok-lee/ecotyper-py.git

and install with pip.

$ pip install .

Usage

To run EcoTyper pipelines, instantiate an EcoTyper object and run either discovery() or recovery(). Recovery is not yet implemented.

eco = EcoTyper(adata)
eco.discovery(
    cell_type_idx,
    sample_idx,

    n_jobs              = 10,
    fractions           = "cell_type_specific",
    n_cells_gene_filter = 500,
    n_cells_distance    = 2500,
    n_genes_top_nmf     = 1000,
    cophenetic_cutoff   = 0.95,
    max_clusters        = 20,
    nmf_restarts        = 5,
    random_state        = 0,

    figures             = "output/folder"
)