DrugSynergyEngine

Introduction

Drug combination therapy exploits synergistic interactions to enhance efficacy, reduce toxicity, and overcome resistance. Systematic computational analysis of drug combination dose-response matrices enables identification of synergistic pairs from high-throughput screens. We present DrugSynergyEngine, implementing three complementary synergy models.

Methods

Hill Equation Fitting

Single-agent dose-response curves are fit to the Hill (sigmoidal) equation: E(d) = Emin + (Emax - Emin) / (1 + (IC50/d)^n)

Parameters (IC50, Emax, Hill coefficient n) estimated by scipy.optimize.curve_fit with bounds constraints.

Bliss Independence Model

Expected additive response for combination (dA, dB): E_Bliss(dA, dB) = E(dA) + E(dB) - E(dA)×E(dB)

Bliss synergy score = mean(E_observed - E_Bliss) across the dose matrix.

Loewe Additivity Model

Based on dose equivalence: combination index CI = dA/IC50_A_eff + dB/IC50_B_eff. CI < 1: synergy; CI = 1: additivity; CI > 1: antagonism.

HSA (Highest Single Agent) Model

E_HSA(dA, dB) = max(E(dA), E(dB)) HSA synergy = mean(E_observed - E_HSA).

Synergy Landscape

2D heatmaps and 3D surface plots of synergy scores across the dose matrix. Isobologram analysis for selected pairs.

Results

• 10 drugs, 20 combination pairs, 8×8 dose matrices
• Hill equation mean R²=0.997 (excellent fit)
• 4 synergistic pairs (Bliss score > 8):
  1. Olaparib+Venetoclax (Bliss=9.5, PARP+BCL2, AML)
  2. Ibrutinib+Venetoclax (Bliss=8.9, BTK+BCL2, CLL)
  3. Palbociclib+Fulvestrant (Bliss=8.3, CDK4/6+ER, breast)
  4. Osimertinib+Bevacizumab (Bliss=8.1, EGFR+VEGF, NSCLC)

Conclusion

DrugSynergyEngine provides a complete, executable drug synergy analysis pipeline identifying clinically validated combinations with high accuracy.

Code

https://github.com/junior1p/DrugSynergyEngine

pip install numpy scipy pandas matplotlib
python drug_synergy_engine.py