Tutorial.md

This tutorial is designed to walk through a basic example of network inference in Yeast and the basic mechanism for constructing an inference workflow for an arbitrary data set

Set Up the Inferelator from the GitHub repository

Clone the codebase:

git clone https://github.com/flatironinstitute/inferelator.git

Enter its top-level directory:

cd inferelator

Install required python libraries:

python -m pip install -r requirements.txt

Install required libraries for parallelization (running on a single machine requires only python -m pip install pathos):

python -m pip install -r requirements-multiprocessing.txt

Install the inferelator package

python setup.py develop --user

Set Up the Inferelator using pip

Install the inferelator

python -m pip install inferelator --user

Download the example data files and run scripts from Zenodo

wget https://zenodo.org/record/3355524/files/inferelator_example_data.tar.gz

Unpack the example data files and run scripts

tar -xzf inferelator_example_data.tar.gz

Change the value of DATA_DIR = in the example scripts to point to the just-created example data paths

Run network inference on Yeast Microarray data

python examples/yeast_network_inference_run_script.py

Acquire necessary data for network inference in a different organism

Obtain expression data and save it as a TSV file of [Samples x Genes] (Other file formats are also possible; AnnData h5 .h5ad files are recommended for large or sparse data)

Obtain prior interaction data between TFs and target genes and save it as a TSV file of [Genes x TFs]

Create a list of TFs to model for inference and save it as a text file with each TF on a separate line [TFs]

Note that the TF and Gene names must match between files, and these files can be compressed with gzip if needed

Construct a new run script (new_organism.py) for a different organism

Select parallelization options:

from inferelator.distributed.inferelator_mp import MPControl

# The if __name__ == '__main__' pragma prevents runaway spawning
# when os.fork is not available. 

if __name__ == '__main__':
    MPControl.set_multiprocess_engine("multiprocessing")
    MPControl.client.processes = 4
    MPControl.connect()

Create an inferelator workflow

from inferelator import workflow
worker = workflow.inferelator_workflow(regression="bbsr", workflow="tfa")

Set file names and paths:

worker.set_file_paths(input_dir='data/new_organism',
                      output_dir='~/new_organism/',
                      tf_names_file='tf_list.txt',
                      priors_file='new_prior.tsv.gz',
                      gold_standard_file='new_prior.tsv.gz')
worker.set_expression_file(tsv='new_expression.tsv.gz')

Add a line to execute inference:

worker.run()

This script can now be run from the command line as python new_organism.py