Research

We combine techniques from combinatorial optimisation and machine learning to develop accurate and efficient computational methods. For example, we have developed exact and approximate algorithms for a graph colouring problem to increase the resolution of experimental protein structure data, used neural networks to project high-dimensional cellular measurements into an interpretable low-dimensional space, and extended dynamic time warping to compare complex trajectories, e.g. of differentiating immune cells.

The software tools we have developed address important biological and medical issues. For example, in close collaboration with biologists and clinicians, we have contributed to the discovery of the embryonic origin of adult neuronal progenitor cells and established a link between TIM-3 expression and an increased risk of relapse in paediatric patients with acute lymphoblastic leukaemia.