How do cells with the same genetic material build systems that vary in form and function? This question lies at the heart of developmental biology in mammals. Our hypothesis is that mammalian cells contain gene programs that naturally give rise to variation within cell populations. These gene programs may be exploited by cancers, driving tumor heterogeneity and treatment failure.
We test these ideas with a combination of bioinformatics, machine learning, cell biology, genomics, and technology development.
Many genes show highly heterogeneous expression across populations of cells, with high expression in some cells and low expression in other cells
The genome contains the instructions for many different outcomes for each cell. We are studying how transcription factors and enhancers interact to determine these outcomes.
In the physical world, differences in cell types evolved over billions of years. Now, in the virtual world, we can replicate the process in hours using tools from artificial intelligence.
Despite decades of study, the most irreducible principles of cancer remain elusive. One idea we are pursuing is that activation of developmental programs in the wrong time and place drives tumors and tumor heterogeneity. We are employing 'big data' approaches, including machine learning and A.I. alongside wetlab experiments, to identify new therapeutic targets in cancer.
