Dr. Cen Wu
In cancer research, profiling studies have been extensively carried out to obtain multiple types of genomic measurements such as mRNA expression levels, copy number variations, DNA methylation and histone modifications, and many others. Many existing studies are “one dimensional” and focus on one type of genomic measurement. More recently, profiling the same samples on diverse layers of genomic activities has made it possible to borrow strength from the cancer genomics data across multiple platforms and conduct the “multi-dimensional” studies.
My research is to develop novel and efficient statistical machine learning methods for the integrative analysis of multiple types of cancer genomic data, in order to better elucidate cancer etiology and prognosis. Such analysis is challenging in that the dimension of measurements for even single category of genomic data is much larger than the sample size, and many conventional statistical approaches fail to work well. Our methods are particularly powerful for analyzing such high dimensional data and can be extended to multiple types of cancers. Our study will lead to practically useful models under cancer outcomes and more accurate identification of important prognostic markers for clinical applications.