The success rate of cancer clinical trials presently stands below 5%. Latest studies indicate that use of biomarkers to stratify patients entering clinical trial can improve success rate up to 2 times. Traditionally, a limited number of biomarkers (1 to 3) is used to identify responsive patients. Our approach will allow integrating information about the status of thousands of DNA and RNA markers to predict patient response with high accuracy using large scale mathematical models of cell type specific signalling networks in cancer. Other model applications include prediction of drug resistance mechanisms and new target identification.