meK-Means: Biophysically Motivated and Interpretable Inference of Cell Types from Multimodal Sequencing Data

Notebooks for reproducing all figures and analysis of simulated and single-cell datasets for the meK-Means paper. All saved/processed data used for analysis can be found on CaltechData. Figure created with BioRender.com.

Getting Started

------ For a tutorial of how to use meK-Means -----

See the example_meKMeans_notebook.ipynb

------ For a tutorial in generating U and S counts -----

See the get_data_example_notebook.ipynb.

For notebooks that run on Google Colab, you will see the Colab link included at the top of the notebook. Just click on the symbol.

An introduction to using Google Colab can be found here. Briefly, run each code cell by selecting the cell and executing Command/Ctrl+Enter.

#To install meK-Means

pip install monod

import monod
from monod import mminference #Function implementing meK-Means algorithm

Related Links

meK-Means utilizes the Monod package for single-cell, CME-based inference.

Repo Directory Contents

1. example_meKMeans_notebook.ipynb: Tutorial notebook for using meK-Means to cluster single-cell RNA-seq data.

2. get_data_example_notebook.ipynb: Tutorial notebook of how to obtain U and S counts from single-cell RNA-seq FASTQs.

3. analysis_notebooks All analysis notebooks from which the paper figures were generated. Each can be run standalone.

   Main Figure Notebooks :

   • Fig1_standard.ipynb: Notebook for generating Fig. 1 standard clustering plots.
   • Figs_2_3_plots.ipynb: Notebook for generating Fig. 2 and 3 benchmark results plots.
   • Fig4_explorData.ipynb: Notebook for generating Fig. 4 exploratory analysis plots.

   Preprocessing Notebooks listed in order of operation, to generate U/S counts and simulations:

   1. sim_benchmark_data_gen/alldataset_preprocess.ipynb: Processes fastqs for all datasets used, into U/S counts.
   2. sim_benchmark_data_gen/gen_benchmark_data.ipynb: Processes alldataset_preprocess to select genes and save data for benchmarking.
   3. sim_benchmark_data_gen/gen_sims.ipynb: Generates simulated, clustered data based on biophysical model of transcription.

   Other Analysis Notebooks:

   • sim_benchmark_meK_Leiden_KMeans_GLUE: Notebook for running meK-Means, Leiden, K-Means, GLUE on all benchmark and simulated data.
   • sim_benchmark_WNN.ipynb: Notebook for running Seurat WNN and MOFA+ on all benchmark and simulated data.
   • sim_benchmark_scMDC_scVI.ipynb: Notebook for running scVI and scMDC on all benchmark and simulated data (uses Colab GPU).
   • Supp_dropout.ipynb: Notebook for analysis of meK-Means with simulated dropout data.
   • Supp_timing.ipynb: Notebook for runtime benchmarking of meK-Means.

4. analysis_output

   • Saved result files (clustering method results) from analysis notebooks.

5. scripts

   • Python script to extract germ cell dataset metadata and scMixology data metadata.

6. env

   • Python Conda environment (py39.yml) for Linux.
   • R Conda environment (r_Seurat.yml) for Linux.