There is an enormous current interest in identifying genetic determinants of interindividual differences in the efficacy and toxicity of cancer medications and in tailoring treatment regimens to each patient's genomic profile. The volume and complexity of data from these pharmacogenomic studies and individualized therapy trials pose unique statistical and computational challenges. The broad, long-term objectives of this project are to develop novel and high-impact statistical methods and computational tools for the designs and analysis of such cancer studies.
We will focus on four specific aims:
Aim 1. Develop statistical methods to assess the impact of DNA variations on the occurrence of adverse clinical events (e.g., neuropathy, neutropenia, and hypertension) in cancer clinical trials under complex censoring and sampling schemes.
Aim 2. Develop statistical methods for integrative analysis of multi-platform genomics data (DNA variations, methylation levels, RNA expressions, protein expressions, etc.) to elucidate the genomic influence on treatment response.
Aim 3. Develop statistical methods to discern tumor cell subclones and to relate intra-tumor heterogeneity to clinical outcomes.
Aim 4. Develop new machine learning techniques to discover and validate biomarkers.