COSIFER - COnSensus Interaction Network InFErence Service

Inference framework for reconstructing networks using a consensus approach between multiple methods and data sources.

The online documentation for the module can be found here.

requirements

• Python >= 3.6
• R >= 3.6.0
• pip
• wget
• build-essential (make, g++, cmake)
• git
• libgit2

installation

Install it using pip to use it as module:

pip install git+https://github.com/PhosphorylatedRabbits/cosifer@master

Or install it via pip after cloning (suggested for development):

cd cosifer
pip install -U -e .

Note for Mac OS X

To install some of the requirements a compiler supporting OpenMP is needed. We suggest to use [brew] to install it and use it for the compilation as follows:

brew install gcc  # this might take time
CC=/usr/local/bin/g++-9 pip install git+https://github.com/PhosphorylatedRabbits/cosifer@master

COSIFER executable script

The COSIFER executable accept the following arguments:

usage: cosifer [-h] -i FILEPATH -o OUTPUT_DIRECTORY [--standardize]
               [--no-standardize] [--samples_on_rows] [--sep SEP]
               [--fillna FILLNA] [--header HEADER] [--index INDEX]
               [--methods METHODS] [--combiner COMBINER]
               [--gmt_filepath GMT_FILEPATH]

Run COSIFER to perform network inference on given data.

optional arguments:
  -h, --help            show this help message and exit
  -i FILEPATH, --filepath FILEPATH
                        path to the data.
  -o OUTPUT_DIRECTORY, --output_directory OUTPUT_DIRECTORY
                        path to the output directory.
  --standardize         flag that indicates whether to perform
                        standardization. Defaults behaviour is to standardize
                        the data.
  --no-standardize      flag that indicates whether to suppress
                        standardization. Defaults behaviour is to standardize
                        the data.
  --samples_on_rows     flag that indicates that data contain the samples on
                        rows. Defaults to False.
  --sep SEP             separator for the data. Defaults to .
  --fillna FILLNA       fill NAs with a given value. Defaults to 0..
  --header HEADER       header index in the data. Defaults to 0.
  --index INDEX         column index in the data. Defaults to None, a.k.a., no
                        index.
  --methods METHODS     methods which should be used for network inference. If
                        no methods are passed, it defaults to the following
                        set of methods: ['pearson', 'spearman', 'aracne',
                        'mrnet', 'clr', 'funchisq']
  --combiner COMBINER   method used for consensus network inference. Defaults
                        to: summa
  --gmt_filepath GMT_FILEPATH
                        optional GMT file to perform inference on multiple
                        gene sets.

examples

In the folder examples, you can find applications of COSIFER consensus network inference.

web service deprecation

The COSIFER web-service has been deprecated in favour of the python package hosted in this repository. Please check the examples to use COSIFER in your research.

citation

If you use COSIFER in you research please cite:

@article{10.1093/bioinformatics/btaa942,
    author = {Manica, Matteo and Bunne, Charlotte and Mathis, Roland and Cadow, Joris and Ahsen, Mehmet Eren and Stolovitzky, Gustavo A and Martínez, María Rodríguez},
    title = "{COSIFER: a python package for the consensus inference of molecular interaction networks}",
    journal = {Bioinformatics},
    year = {2020},
    month = {11},
    issn = {1367-4803},
    doi = {10.1093/bioinformatics/btaa942},
    url = {https://doi.org/10.1093/bioinformatics/btaa942},
    note = {btaa942},
    eprint = {https://academic.oup.com/bioinformatics/advance-article-pdf/doi/10.1093/bioinformatics/btaa942/34088187/btaa942.pdf},
}