Dr. Wang's research is motivated by an interest in large-scale complex data sets in recent genomic and familial studies and by important biological questions that emerge from the analysis of these data. Her current research focuses on 1) the development of methods and software for accurate measurement of high-throughput genomic data, and 2) the development and validation of statistical approaches and software for personalized cancer risk prediction.

It is not an easy task to extract genomic information of interest from the raw signals that come directly from chemical or physical reactions. Current high-throughput technologies have all inevitably incorporated multi-level confounders that affect the observed signals. The large amount of data they produce also make it difficult to calibrate these technologies using "gold standards" that are usually generated by experiments that are more accurate but are low-throughput and expensive. Dr. Wang's work focuses on using statistical modeling to make the interpretation of raw high-throughput signals more accurate. Her group has been working on methods and tools for the accurate calling of somatic mutations in the sequencing data from matched pairs of tumor-normal samples, as well as for the deconvolution of mixed transcriptomes in heterogeneous tumor samples.

Cancer results from the accumulation of multiple genetic mutations. A germline mutation in a cancer gene predisposes the carrier to the development of cancer. This "inherited susceptibility" results in the familial clustering of cancers, known as "familial cancer syndromes." Mendelian carrier probability models based on Bayesian methods that use detailed family history as input can help clinical researchers identify cancer patients at earlier and more treatable stages and identify healthy individuals at high risk of developing cancer. Dr. Wang's work in this field focuses on 1) applying Mendelian models to cancers of interest for personalized risk assessment, and 2) developing methodologies for the evaluation of risk assessment models using family and correlated data.