2017 Joaquim Aguirre-Plans

Structural Bioinformatics Laboratory

Universitat Pompeu Fabra

Drug Interactions ANAlyzer (DIANA) is a Python (v3) software which permits the prediction of drug combination using drug target and protein-protein interaction information

• scripts: Folder where the scripts to run the program are stored
• scripts/generate_profiles.py: Script to generate the profiles of a drug
• scripts/compare_profiles.py: Script to compare the profiles of two drugs
• diana: Folder where the source code is stored
• workspace: Folder where the results of the analysis are stored
• workspace/profiles: Folder containing the profiles generated
• workspace/comparisons: Folder containing the comparisons of profiles
• config.ini: Configuration file, containing the default parameters to run DIANA

Before starting running DIANA, we need to modify the parameters in the configuration file as we want.

The configuration file is in the main directory, as config.ini

Here we can modify the parameters of BIANA database if we have it.

If we do not have BIANA, we can set use_cluster as True and use the downloaded data.

We also have to introduce the path to Guild executable. In case that you do not have it, you can download it and install it from https://github.com/emreg00/guild

Now, create the profiles of the drugs that you want to compare. To create the profile of one drug, run the following script:

python path/to/diana/scripts/generate_profiles.py 
  -j <job id> # Give an specific job identifier. If not, it will be created automatically
  -d <drug name>
  -t <drug targets (in Entrez geneID)>
  -pt <type IDs of the proteins (default is geneid)>
  -sif <network (in SIF format and geneID)>
  -th <thresholds list> # Cut-offs that will define the profiles generated
  -ws <workspace> # Directory where the results will be saved

Example of generate the profile having network and the targets:

python path/to/diana/scripts/generate_profiles.py -j metformin_with_targets -d metformin -t path/to/diana/workspace/targets/metformin.targets -sif path/to/diana/workspace/sif/human_eAFF_geneid_2017.sif

Example of generate the profile having network but not having targets. The targets will be searched in BIANA.

python path/to/diana/scripts/generate_profiles.py -j metformin -d metformin -sif path/to/diana/workspace/sif/human_eAFF_geneid_2017.sif

With this script we can compare the profiles of a pair of drugs:

python path/to/diana/scripts/compare_profiles.py 
  -j1 <job 1>
  -j2 <job 2>
  -sif <network (in SIF format and geneID)>
  -th <thresholds list> # Cut-offs that will define the profiles generated
  -ws <workspace> # Directory where the results will be saved

In the following example, we compare the profiles created for metformin with the profiles created for haloperidol:

python path/to/diana/scripts/compare_profiles.py -j1 metformin -j2 haloperidol -sif path/to/diana/workspace/sif/human_eAFF_geneid_2017.sif

There is a pipeline on how to analyze results in README_analysis_one_target.