The Self-Assembling-Manifold algorithm


Keywords
scrnaseq, analysis, manifold, reconstruction
License
MIT
Install
pip install sam-algorithm==0.8.4

Documentation

Build Status

self-assembling-manifold -- SAM version 0.8.4

The Self-Assembling-Manifold (SAM) algorithm.

SAM is now on Scanpy!

https://github.com/theislab/scanpy

import scanpy as sc
import scanpy.external as sce

#returns the SAM object if `inplace=True` and (SAM,AnnData) otherwise
sam = sce.tl.sam(adata, inplace=True) #adata is your AnnData object

Requirements

  • numpy
  • scipy
  • pandas
  • scikit-learn
  • umap-learn
  • numba
  • anndata

Optional dependencies

  • Interactive GUI (Jupyter notebooks)

    • plotly==4.0.0
    • ipythonwidgets
    • jupyter
    • colorlover
    • ipyevents
  • Plots

    • matplotlib
  • Clustering

    • louvain
    • leidenalg
    • hdbscan
    • cython
  • scanpy

Installation

Docker

Build the Docker image with:

git clone https://github.com/atarashansky/self-assembling-manifold.git
cd Docker
bash build_image.sh

Run the Docker image with:

bash run_image.sh

It will ask you to provide the image name, container name, port to run the Jupyter notebook server on, and the path to a directory that will be mounted onto the Docker container's file system.

Anaconda

SAM requires python>=3.7. Python can be installed using Anaconda.

Download Anaconda from here: https://www.anaconda.com/download/

Create and activate a new environment with python3.7 as follows:

conda create -n environment_name python=3.7
conda activate environment_name

Having activated the environment, SAM can be downloaded from the PyPI repository using pip or, for the development version, downloaded from the github directly.

PIP install:

pip install sam-algorithm

Development version install:

git clone https://github.com/atarashansky/self-assembling-manifold.git
cd self-assembling-manifold
python setup.py install

For plotting, install matplotlib:

pip install matplotlib

For interactive data exploration (in the SAMGUI.py module), jupyter, ipythonwidgets, colorlover, ipyevents, and plotly are required. Install them in the previously made environment like so:

conda install -c conda-forge -c plotly jupyter ipywidgets plotly=4.0.0 colorlover ipyevents

Enabling the SAM GUI in JupyterLab

If you use Jupyter Notebooks, these steps are not needed. If you would like to be able to run SAMGUI in JupyterLab, please do the following:

First install nodejs with: conda install nodejs

To enable ipythonwidgets in Jupyter lab, please run the following:

jupyter labextension install @jupyter-widgets/jupyterlab-manager@1.0 --no-build
jupyter labextension install plotlywidget@1.1.0 --no-build
jupyter labextension install jupyterlab-plotly@1.1.0 --no-build
jupyter lab build

SAMGUI should now work in JupyterLab.

Running the SAM GUI

The SAM GUI interface can be run in Jupyer notebooks with the following:

from samalg.gui import SAMGUI
sam_gui = SAMGUI(sam) # sam is your SAM object
sam_gui.SamPlot

Please see the plotting tutorial for more information about the GUI interface.

SAM GUI example image

Tutorial

Please see the Jupyter notebooks in the 'tutorial' folder for basic tutorials. If you installed a fresh environment, do not forget to install jupyter into that environment! Please run

pip install jupyter

in your conda environment. The tutorial assumes that all optional dependencies are installed.

Basic usage

There are a number of different ways to load data into the SAM object.

Using the SAM constructor

Using preloaded scipy.sparse or numpy expression matrix, gene IDs, and cell IDs:

from samalg import SAM #import SAM
sam=SAM(counts=(matrix,geneIDs,cellIDs))
sam.preprocess_data() # log transforms and filters the data
sam.run() #run with default parameters
sam.scatter()

Using preloaded pandas.DataFrame (cells x genes):

from samalg import SAM #import SAM
sam=SAM(counts=dataframe)
sam.preprocess_data() # log transforms and filters the data
sam.run() #run with default parameters
sam.scatter()

Using an existing AnnData object:

from samalg import SAM #import SAM
sam=SAM(counts=adata)
sam.preprocess_data() # log transforms and filters the data
sam.run() #run with default parameters
sam.scatter()

Using the load_data function

Loading data from a tabular file (e.g. csv or txt):

from samalg import SAM #import SAM
sam=SAM() #initialize SAM object
sam.load_data('/path/to/expression_data_file.csv') #load data from a csv file
#sam.load_data('/path/to/expression_data_file.txt', sep='\t') #load data from a txt file with tab delimiters
sam.preprocess_data() # log transforms and filters the data
sam.load_annotations('/path/to/annotations_file.csv')
sam.run()
sam.scatter()

Loading an existing AnnData h5ad file:

If loading tabular data (e.g. from a csv), load_data by default saves the sparse data structure to a h5ad file in the same location as the tabular file for faster loading in subsequent analyses. This file can be loaded as:

from samalg import SAM #import SAM
sam=SAM() #initialize SAM object
sam.load_data('/path/to/h5ad_file.h5ad') #load data from a h5ad file
sam.preprocess_data() # log transforms and filters the data
sam.run()
sam.scatter()

Saving/Loading SAM

If you wish to save the SAM outputs and raw and filtered data, you can write sam.adata to a h5ad file as follows: sam.save_anndata(filename).

You can load this data back with sam.load_data: sam.load_data(filename)

Citation

If using the SAM algorithm, please cite the following eLife paper: https://elifesciences.org/articles/48994

Tarashansky, A. J. et al. Self-assembling manifolds in single-cell RNA sequencing data. eLife 8, e48994 (2019).

Adding extra functionality

As always, please submit a new issue if you would like to see any functionalities / convenience functions / etc added.