The execution of mitosis requires that the activity of protein complexes be tightly coordinated in space and time but our knowledge of the dynamics and interactions of the several hundred proteins required for mitosis is incomplete and fragmented due to lack of systematic and quantitative approaches. In addition, the few existing models only offer limited insight into oversimplified and isolated aspects of mitosis. To tackle this situation, our group uses automated fluorescence imaging of live dividing cells, image analysis and machine learning approaches to build a canonical model of a human dividing cell that can be used to integrate data from different experiments. The simultaneous visualization of multiple proteins in the model cell can be used to generate new testable hypotheses about protein functions in mitosis.
National Institute of Standards and Technology