Our group specializes in methods for the analysis and interpretation of high-throughput measurements with the goal to understand mechanisms of tumor development and drug resistance
Predictive biomarkers of agent-specific drug response in cancer patients
For the most common cancers of the breast, lung, or colon, an individual patient has a 20-40% chance of responding to a particular chemotherapeutic agent. Thus, to maximize the chances of curing the patient, a typical treatment regimen may involve several drugs, most of which are not effective against the cancer and cause only side effects. We use causative learning approaches such as RNAi screens to discover candidate predictive biomarkers of drug response, which we then evaluate in tumor molecular (RNA, DNA) profiles from clinical trials. These predictive biomarkers will enable personalized patient treatment and reduce unnecessary suffering caused by ineffective therapy.
Novel methods for molecular characterization of tumors
A typical gene expression or CGH microarray measurement yields 104 ~ 106 values per array. We are developing methods to reduce this data to a small number of robust, biologically- or clinically-relevant parameters. These parameters will allow us to probe the underlying status of a given tumor, which may prove to be prognostic or predictive of drug response.