Predictive modelling of drug treatment responses

Comprehensive testing of the sensitivity of patient-derived cancer samples to a wide collection of chemical compounds is increasingly being used to tailor drug treatments for individual cancer patients. However, high-throughput drug testing experiments result in high-dimensional datasets, with inherent measurement noise and technical variability, which hinders many downstream analyses, such as detection of differential drug sensitivities or stratification of patients based on their selective response patterns. This talk describes the computational analysis pipeline developed and used at FIMM, which enables reliable quantification of differential drug sensitivities, mapping of target addiction networks behind the individual response profiles, prediction of targeted drug combinations for relapsed patients, as well as identification of biomarkers predictive of selective drug responses. This experimental-computational approach is illustrated in both hematological cancers as well as in solid tumors.