Project abstract for group vanness

Pharmacogenomics in Cancer

Variation in drug response is a serious concern in achieving desirable response in the treatment of cancers. These researchers have successfully developed a prediction algorithm for transcriptional classification of drug response (Transcript CDP) that uses gene expression profile (GEP) signatures to predict response and resistance to proteasome inhibitors in myeloma. Their plan is to apply this approach to build classification models and generate drug response classifiers (feature vectors/GEP signature) using the cell lines, gene expression analysis and cytotoxic response data for a wider array of drugs. The drugs screened include a wide spectrum of targets and processes involved in cancers, including targeted agents and cytotoxic chemotherapeutics that are already approved for clinics, drugs currently in various phases of clinical trials, and experimental tool compounds. The predictive programming uses a system the lab developed that builds classification models, performs feature selection, and predicts PI response using an ensemble forecasting algorithm that uses a combination of LASSO, Random Forest, and Support Vector machine learning methods.

Return to this PI's main page.