🧬 TRAFIKK

Systematic prediction and mechanistic interpretation of anticancer drug synergies

---

Trafikk is a computational pipeline for in silico prediction and mechanistic interpretation of drug combination responses in cancer. It integrates cell-line-specific molecular contexts with Boolean network modelling and signal-propagation analysis to identify synergistic drug pairs and explain how synergy emerges at the pathway level.

⚙️ Pipeline

Trafikk simulates drug perturbations on cell-line-calibrated Boolean models of cancer signalling networks, generating functional response profiles for single drugs and combinations. These profiles enable both synergy classification and mechanistic interpretation of the underlying signalling dynamics.

 Celios ➜ Gitsbe ➜ Drexpa ➜ Oris ➜ Synco ➜ Siflex
   │         │         │        │       │        │
 Omics    Models    Drugs    Synergy  Bench   Analysis

Modules

	Module	What it does	Language
🧬	Celios	Integrates cell-line omics data (mutations, CNV, TF activity) to calibrate the base network to specific biological contexts	Python
🔧	Gitsbe	Generates ensembles of logic-based models for each calibrated cell-line network	Java
💊	Drexpa	Maps experimental drug panels to in silico perturbation profiles using public target databases (GDSC, OpenTargets, ChEMBL, UniProt, BindingDB)	Python
⚡	Oris	Computes in silico viability and synergy scores via signal-propagation analysis (built on BooLEVARD)	Python · HPC
📊	Synco	Benchmarks predictions against experimental synergy data using standard classification metrics (AUC-ROC, AUC-PR, F1, accuracy, recall, precision)	Python
🔬	Siflex	Performs pathway-level functional analysis of drug effects and generates mechanistic hypotheses for synergistic responses	Python

_{Celios, Drexpa, Synco — V. Bermúdez · Gitsbe — J. Zobolas · Oris — M. Fariñas · Siflex — M. Fariñas, V. Bermúdez}

🚀 Installation

Each module is installed independently from its own repository:

# Python modules
pip install git+https://github.com/druglogics/celios.git
pip install git+https://github.com/druglogics/drexpa.git
pip install git+https://github.com/druglogics/oris.git
pip install git+https://github.com/druglogics/siflex.git

# Synco (notebook-based)
pip install git+https://github.com/ViviamSB/SYNCO.git

# Gitsbe — see its repo for Java build instructions
# https://github.com/druglogics/gitsbe

Refer to each module's repository for detailed dependency and environment requirements.

📖 Documentation

Full unified documentation is available at druglogics.github.io/trafikk.

🧪 Synergy quantification

Drug synergy is assessed using Bliss independence:

$$\Delta_{\text{Bliss}} = V_{AB} - V_A \cdot V_B$$

where $V_{AB}$ is the viability under combined perturbation and $V_A$, $V_B$ are the single-drug viabilities. Negative values indicate synergy.

🏗️ Built upon

Project	Role
DrugLogics	Model generation and calibration
BooLEVARD	Signal-propagation analysis in Boolean models

📝 Citation

Fariñas M.*, Bermúdez V.*, Tsirvouli E., Lippestad K., Zobolas J., Aittokallio T., Lehti K.†, Flobak Å.† TRAFIKK: systematic prediction and mechanistic interpretation of anticancer drug synergies. Submitted.

📄 License

This project is licensed under the GNU General Public License v3.0.