Cancer cell populations within a tumor are often diverse, and the dynamic shifting of cell state plays a major role in treatment resistance and cancer metastasis. Dr. Copperman is utilizing tools from statistical physics and modern machine learning to predict how subpopulations of cancer cells continuously adapt to survive and eventually metastasize to other organs in the body. Using time-resolved single-cell imaging and molecular measurements to inform data-driven models, he aims to develop predictive whole-cell modeling and optimal control strategies for heterogeneous cellular populations.

Dr. Copperman performs image processing and single-cell analysis using python script. He is also developing single-cell analysis methods (featurization, tracking, trajectory embedding, integration with sequencing data) in a shareable open-source python implementation. Other computational methods include integrating non-equilibrium statistical physics concepts with unsupervised clustering / manifold learning, global optimization, markov- chain monte carlo sampling, deep learning and neural-network application, and linearization and matrix methods.