dc.description.abstract | In the field of high-throughput drug screening, particularly in relation to drug combinations, synergy scores have traditionally dominated as the most used method to determine the efficacy of drug pairs. However, for decades, synergy scores have demonstrated a lack of concordance and are particularly vulnerable to outliers. To address these challenges, this thesis introduces concepts from causal inference, such as counterfactual notation, and applies them to express drug synergy. This representation offers clarity and precision, providing a better foundation to understand the concepts of synergy. Given the increased interest and utilization of drug combination datasets, we argue for a paradigm shift: rather than relying on synergy models, we should leverage direct comparison of drug combinations for a more straightforward method. This will also combat the issue of synergy scores that can lead to contradictory results, as has been previously documented. Focusing on in vitro oncology cell line experiments, where the primary objective is inhibition, we propose the assumption that the dose-response surface is inherently monotone. To ensure the robustness of this assumption, we introduce statistical tests to assess the monotonicity of the dose-response curve. When monotonicity is confirmed, the study uses isotonic regression, a non-parametric method with minimal assumptions also robust to outliers, to model the dose-response relationship. Our approach paves the way for more direct, reliable, and intuitive assessments of drug combinations, particularly when utilizing known or approved drugs. | |