Gene expression and regulation is a complex process involving a multi-layer process. Many studies have focused on predicting gene expression from DNA accessibility and the underlying sequence, mostly through sophisticated deep learning models. In our lab, we would like to push this boundary, and infer expression from plasma cell-free DNA.

Nearly half of non-small cell lung cancer patients are diagnosed with localized disease, making them candidates for chemoradiotherapy (CRT) as curative treatment. Among these, CRT is the main treatment for stage III patients, of whom, ~30% will experience local recurrence. Despite advances utilizing tumor imaging techniques to improve targeting, no further personalization is done based on tumor biology to refine radiation strategy. Therefore, relevant biomarkers to predict treatment outcome to guide and adapt CRT are lacking. In our lab, we address this problem by profiling cell-free DNA using state-of-the-art techniques to infer RNA expression from DNA, and thereby identify determinants of response to CRT.

Our laboratory focuses on developing techniques for reliable genotyping and phenotyping for better disease characterizations (e.g. classic Hodgkin lymphoma, and small cell lung cancer). Towards this goal, we will develop new tools for detecting various structural variations, single and phased variants, single cell sequencing of RNA, ATAC, and DNA.

Our laboratory is also interested in developing methods to interpret and analyze longitudinal data from multi-omics. We’re in particular interested in multi-scale time series modeling for (1) detecting emergent lesions that lead to resistance to various treatments, and (2) decision making. To this end, we are collaborating with various laboratories to collect time-series from cancer patients.