Installation and Usage

Install

Prerequisites: maven 3.6.0 and Java 8.

First, install Gitsbe. Then:

git clone https://github.com/druglogics/drabme.git
cd drabme
mvn clean install

The above command creates a package <name>-jar-with-dependencies.jar file with all dependencies installed, in the target directory.

Alternatively, you could just use directly one of the released drabme packages.

Note that Drabme calculates attractors for the boolean models it generates using either the BNReduction tool (Veliz-Cuba et al. 2014), the BioLQM Java library (Naldi 2018) or the Most Permissive Boolean Networks framework (Paulevé et al. 2020). BioLQM is included by default in the code. The BNReduction and MPBN scripts have to be manually installed following the respective documentation.

Example

The recommended way to run Drabme is to use it’s Launcher. From the root directory of the repo run:

cd example_run_ags
java -cp ../target/drabme-1.2.1-jar-with-dependencies.jar eu.druglogics.drabme.Launcher --project=test --modelsDir=models --drugs=toy_ags_drugpanel.tab --perturbations=toy_ags_perturbations.tab --config=toy_ags_config.tab --modeloutputs=toy_ags_modeloutputs.tab

or run the mvn profile directly (same input as the command above through the pom.xml):

mvn compile -P runExampleAGS

Inputs

Running the Drabme Launcher with no parameters, generates a usage message with the available options. The required parameters are:

• --modelsDir: A models directory with files that were generated by Gitsbe (the files are boolean models in .gitsbe format)
• --drugs: drug panel file
• --modeloutputs: model outputs file
• --config: configuration file

The non-required parameters are:

• --project: the project name which is used as the name of the directory where the outputs will be stored.
• --perturbations: perturbations file

Outputs

The expected outputs of Drabme’s Launcher can be classified into 3 categories of generated files (see also Description):

1. Model-wise results
   • A model-wise responses file (for every tested perturbation, all models’ global output responses)
   • A model-wise synergies file (for every drug combination tested, the number of models that found it synergistic vs non-synergistic)
2. Ensemble-wise results (or perturbation-wise)
   • A ensemble-wise responses file (for every tested perturbation, the average response of all models that had a global output response - i.e. had an attractor)⁶
   • A ensemble-wise synergies file⁷ (for every drug combination tested, a synergy score indicating how much synergistic that combination was - a more negative number means more synergistic, a positive number indicates antagonism)
3. A model predictions file
   • For every model and drug combination tested, this file stores a value that indicates if the corresponding drug combination manifests synergistically or not in the respective model. The three possible values are: \(0\) = no synergy predicted, \(1\) = synergy predicted, \(NA\) = couldn’t find attractors in either the drug combination inhibited model or in any of the subset-drug inhibited models (this output value is impossible when trapspaces are used for the computation of the attractors).

It is for the better that all input files are on the same directory as in the example above. In any case, the directory with all the output files when running the Launcher will be in the same path as the configuration file.

---

6. This is just the average value of the data in the model-wise responses file↩︎

7. One of the most important outputs of Drabme. The druglogics-roc R shiny app was built to take the results from this output file and produce a ROC & PR curve.↩︎