High-throughput drug sensitivity testing provides a powerful phenotypic
profiling approach to identify effective drug candidates for individual
cell lines or patient-derived samples. Here, we describe an
experimental-computational pipeline, named target addiction scoring
(TAS), which mathematically transforms the drug response profiles into
target addiction signatures, and thereby provides a ranking of potential
therapeutic targets according to their functional importance in a
particular cancer sample. The TAS pipeline makes use of drug
polypharmacology to integrate the drug sensitivity and selectivity
profiles through systems-wide interconnection networks between drugs and
their targets, including both primary protein targets as well as
secondary off-targets. We show how the TAS pipeline enables one to
identify not only single-target addictions but also combinatorial
coaddictions among targets that often underlie synergistic drug
combinations.