Research Interests

Cancer Systems Biology is a mixed experimental and theoretical approach that considers cancer as a complex adaptive system operating over multiple spatial and temporal scales. It can be applied to studying many aspects of cancer progression, including invasion, metastasis, resistance to drugs, effects of mutations, rewiring of signaling networks and response to treatment. It combines the specific effects of altered molecular pathways, genetic networks, growth or differentiation factors, proteases, drugs or other cell-autonomous and microenvironmental perturbations into global system-level quantitative models that may explain unconstrained cell proliferation, the fundamental hallmark of cancer. From these models, the behavior of the system in response to specific perturbations, e.g., drugs, may be simulated and then tested experimentally.

We study melanoma, breast and lung cancer. A special emphasis in our group is quantifying cancer heterogeneity with respect to drug response and to differentiation states. We have developed methods to measure clonal and cell-to-cell variability by high-throughput automated microscopy and image processing. We then derive cancer cell population behavior by integrating single-cell and clonal data.

For example, in one project we aim to define the cellular and molecular variables that drive rebound to therapy in EGFR-oncogene addicted lung cancer. To this end, we subject cancer cells to mutated EGFR targeted drugs, and derive the response at the population level from single-cell data. Hundreds to thousands of single cells or clones are monitored by automated microscopy coupled to image processing and data modeling. We fractionate fates of treated cells into quiescence, apoptosis, or reduced division rate. Experimental data are incorporated into computational models to predict and/or optimize drug response. Model simulations focus on treatment time-course, i.e., depth of response and time to rebound, and make theoretical predictions on specific ways experimental variables may affect cancer cell population responses to treatment. These predictions are validated in vitro in 2D and 3D tissue culture, as well as in vivo with human cell lines xenografted in nude mice or patient derived xenografts. From these data we can also extract molecular determinants that can then be modeled as networks.

In another example, we build transcription factor networks that act as master regulator of differentiation in specific cancer types, and study the relationship between these phenotypic attractor states and response or resistance to treatment.

More generally, the overall goal of our systems approach is to build quantitative hypotheses that translate experimental observations or datasets into computer simulations, based on mathematical modeling techniques including ordinary or partial differential equations, cellular automata, neural networks, immersed boundary method, game theory, Boolean networks. To test these hypotheses, we populate these models with datasets from in vitro or animal experiments, or from clinical material. We then design and perform experiments to validate these predictions, and the outcome of the experimentation is used to evaluate the realism of computer simulations and possibly modify their underlying mathematics. Our group is comprised of an interdisciplinary collection of scientists, including cell and molecular biologists, mathematicians, engineers, bioengineers, bioinformaticians and computational biologists. Our team science thrives on continued personal exchanges among these diverse scientists, looking at cancer research problems through the lenses of different disciplines.