Designing optimal experiments to discriminate interaction graph models

Experiment Planning using Sign Consistency on Influence Graphs

Modern methods for the inference of cellular networks from experimental data often express nondeterminism by proposing an ensemble of candidate models with similar properties. To further discriminate among these model candidates, and to find the biological truth, new experiments need to be carried out. The goal of optimal experiment design is to determine those experiments that discriminate most of the candidates while minimizing the costs. Building upon a notion of consistency between biochemical/genetic regulations and high-throughput profiles of cell activity, ExDesi implements an approach for experiment planning with interaction graph models and sign consistency methods. ExDesi can be used in combination with methods for network inference and consistency checking to compute experiments which are most suitable to deliver results that allow a refinement of the model.


You can install exdesi by running:

$ pip install --user exdesi

On Linux the executable scripts can then be found in ~/.local/bin

and on Mac OS the scripts are under /Users/YOURUSERNAME/Library/Python/3.5/bin.


Typical usage is:

$ exdesi.py candidate_directory experimental_variables_file

For more options you can ask for help as follows:

$ exdesi.py -h
usage: exdesi.py [-h] [--best_set BEST_SET] [-x EXCLUDE]
             networkfiles experivarfile

positional arguments:
  networkfiles          directory of influence graphs in SIF format
  experivarfile         experimental variables

optional arguments:
  -h, --help            show this help message and exit
  --best_set BEST_SET   compute best set of experiments maximal number of
                        experiments, default is OFF, 0=unlimited
  -x EXCLUDE, --exclude EXCLUDE
                        exclude experiments described in file EXCLUDE


Sample files available in the git repository.

A sample call would be:

$ exdesi.py in_silico_HEK293/candidates_round1/ in_silico_HEK293/expvars.txt -x in_silico_HEK293/exclude1.txt


Q: I don’t have pip. How can I install pip without admin rights?

A: To install pip without admin rights:

  1. Download getpip.py.

     $ wget https://raw.github.com/pypa/pip/master/contrib/get-pip.py
  2. Install pip locally.

     $ python get-pip.py --user
  3. You can install using your local pip.

Q: How can I write the output of exdesi into a file?

A: You can redirect the output of exdesi into a file using >. For example to write the results into the file myfile.txt type:

$ exdesi.py network.sif observation.obs --show_labelings 10 --show_predictions > myfile.txt