How can we design CAR T-cell circuits that effectively target tumour cells while minimizing effects on healthy cells? In our new preprint, we approach this as a combinatorial optimization problem. We introduce single- and multi-objective local-search methods for CAR T-cell circuit design and evaluate them on a single-cell cancer atlas. Our multistart local search improves over the previous state-of-the-art approach, while the multi-objective formulation reveals trade-offs between treatment efficacy and safety and can further improve solutions to the constrained design problem.
The work is a collaboration with Bartol Borozan, Alice Descoeudres, Aneta Neumann and Frank Neumann.
Read the preprint on arXiv. (external link, opens in a new window)