Network-based drug effect prediction for precision medicine

Abstract

There has been a growing interest in precision medicine that diagnoses diseases and prescribes drug treatment on the basis of personal profiles. Precision medicine can help improve efficacy of drug treatment and prevent adverse effects. Recently, numerous of computational models have been developed for the advancement of precision medicine system. However, there still are some critical factors that can change drug responses but that have never been considered yet in previous computational models. In this thesis, we study key properties that need to be considered in precision medicine. First, we deal with hormones that can have major changes due to diverse genetic or environmental factors and cause changes in drug responses. Second, we deal with drug resistance that occurs due to diverse causes depending on individuals, and then we discover sensitizers to overcome resistance against anti-cancer drugs. A major strategy for this is to utilize large-scale biological networks to understand biological phenomena from a gradual point of view, from molecular development to cell function. The two models developed in this paper are the first computer models to predict drug efficacy in consideration of hormone or drug resistance. It demonstrates the usefulness of the model by showing outstanding performance in several case studies and by performing cell laboratory validation.