Multi-omics Modeling for Predictive Cancer Immunotherapy

In this seminar, Dr. Elana J. Fertig will demonstrate how CoGAPS, an established method for identifying transcriptional signatures related to cell type and state, can be applied to single cell data to identify patterns underlying immunotherapy response and resistance.

Therapeutic response in cancer is critically dependent on the states of cells (both cancer and immune cells) in the tumor microenvironment, which evolve over time. New single-cell and spatial molecular technologies enable unprecedented characterization of these states across molecular and cellular scales, but they are challenging to interpret because of the multi-dimensional nature of these data. New computational methodologies offer an effective means for interpreting these complex data.

Combining CoGAPS with transfer learning approaches provides insight into the patterns found in preclinical models, which then can be applied to patient samples to find shared features across data sets from different species. This facilitates the discovery of unknown or unanticipated findings, such as the activation of natural killer cells in response to anti-CTLA4 (i.e., an immune checkpoint inhibitor). Spatial distribution of cells in the tumor microenvironment also mediate response and resistance to therapies. Emerging spatial molecular technologies offer a powerful tool for modeling these interactions.