Cancer Biology and Data Science

Our hybrid wet-dry lab views cancer as a whole-body disease – a complex system whose components can be reverse-engineered to guide precision treatment. We apply a wide range of computational and statistical techniques to infer molecular networks underlying cancer using genomic, transcriptomic, and proteomic data. These networks often represent interactions between genes or sets of genes, mediated by a diversity of molecular regulators. We use these networks to generate new hypotheses about how cancer starts, spreads, and responds to treatment. Analyzing human-centric data sets offers a new lens for discovery to advance our understanding of the molecular basis of health and disease. Everyone in our lab works on both cancer biology and data science, in a truly integrated way toward improving cancer detection and care. 

AI-Guided Spatial Biology of Cancer

Spatial biology is a transformative new frontier that enables a realistic view of human disease as it changes constantly at the level of millions of interacting molecules and cells in tissue microneighborhoods. Our lab has been among the first adopters of AI-driven analysis of multimodal spatial biology data to generate human cancer tissue maps as a tumor evolves. Using patient samples acquired throughout disease detection, treatment, and metastasis, we can watch the story of health being told over time.

Systems Biology of Tumor-Immune-Stromal Interactions in Metastatic Progression

Metastasis of cancer cells from the primary tumor to distant sites is the primary cause of cancer-related death. We aim to identify mechanisms of metastasis by which tumor cells instruct immune cells and other stromal cells to tolerate them by focusing on the understudied role of lymph node invasion in tumor-mediated immunosuppression. Our findings promise to provide critical insights into blocking metastatic progression and thereby preventing cancer-related deaths. For more details, see our Center for Cancer Systems Biology website.

Integrating Molecular, Imaging, and Clinical Data

Current projects involve the integration of multi-platform cancer datasets through probabilistic modeling. In a recent collaborative effort through IBIIS, with investigators from the Stanford Departments of Radiology and Surgery, we are creating an association map between CT and PET image features and gene expression microarrays of human non-small cell lung carcinoma. This map illustrates imaging features of lung cancer and enables us to identify prognostic significance image biomarkers by leveraging clinically annotated public data sets.

Mathematically Modeling Cancer Progression to Guide Health Policy

Cancer simulation modeling provides a clinical anchor for our work. We develop multi-scale models of the natural history of cancer that describe the stochastic behavior of tumor growth and metastatic spread. We have used these models to address important health policy questions related to early detection, such as: how does screening mammography and MRI impact breast cancer mortality; and how would CT screening for lung cancer impact lung cancer mortality rates? The findings resulting from this work are used to inform our cancer biology research and also to guide policy aims related to effective lung cancer screening.