SP 3 - DeepLTNBC

Triple negative breast cancer (TNBC) subtypes have distinctive genetic and epigenetic alterations in the DNA damage repair and cell fate response pathways. These alterations are expected to affect the response to drugs. This has been noted for a long time in histological and clinical outcomes, suggesting that treatments should be tailored to specific tumors subtypes. Nevertheless, current treatments use standardized protocols regardless of the cancer subtype. Moreover, there is compelling evidence that the response to chemotherapy also depends on the internal cellular state and timescales of drug response. A vast majority of studies that characterize the response to treatments use bulk population measurements and single time snapshots. In systems composed of uniform cells these population- and time-averaged assays qualitatively reflect the behavior of individual cells and have revealed sophisticated signaling networks. However, tumors contain heterogeneous subpopulations of cells in a dynamically changing environment where inference of single cell identity from bulk approaches is unclear and often not valid. To address these limitations we will  perform comprehensive single-cell studies using both fixed multidimensional snapshots data and live long-term high temporal resolution assays.
In this subproject we will identify and validate metrics to predict responses of a heterogeneous population of cells. We will progress from a static single-timepoint based characterization of cells heterogeneity, to then include spatial information topology metrics, to a dynamic view of cellular states including and lastly a computational model of tumor growth. Aim 1 is to quantify the dynamics of cellular states (e.g. cell cycle, circadian clock) and the time of response to drugs using a subset of TNBC cell lines. Aim 2 is to to exploit the multidimensional immunofluorescence data and patient’s metadata with the use of statistical tools to identify phenotypic states pre- and post-treatment. Aim 3 is to integrate single- cell data with whole tissue organization and its dynamics in a comprehensive computational model of tumor behavior in response to treatment. This model will allow to test hypotheses about therapy outcomes and their molecular determinants. Results will guide new experiments, gain knowledge and consequent refinement of the computational model in an iterative loop between experiments and modelling.