Analysing double-strand breaks in cultured cells for drug screening applications by causal inference

Double strand breaks (DSB) are a hallmark of DNA damage and genetic instability, which are important features of cancer cells. In addition, repair of DSBs provide interesting therapeutic targets. Fluorescence microscopy allows us to visualise DSBs in cells using a dedicated fluorescent marker, which is therefore an informative readout for drug screening applications. We therefore need robust methods in image analysis and statistical analysis to quantify DSBs in single cells and thereby to assess the drug effect with respect to the related pathways. The contribution of this paper is two-fold: first, we compare different DSB quantification schemes; and second we provide a sound statistical framework based on causal inference in order to detect drugs acting directly on DSBs. In particular this second aspect is so far notoriously neglected in the field, even though it is essential for the specific assignment of the drug effect.