Logotyp: till Uppsala universitets webbplats

Glioblastoma (GBM) is the most aggressive form of adult diffuse gliomas, and the most prevalent among malignant primary brain tumors. The prognosis is poor, with a median survival of 15 months and unavoidable tumor recurrence. It is believed that a small population of cells, called cancer stem cells (CSCs), reside within the tumor and are at least partly responsible for the tumor re-initiation. In this thesis, we evaluate a new computational measure, based on the Ising model from statistical physics, for identifying the most stemlike cells in a population by considering single cell RNA-sequencing data. We evaluate a range of different optimizations on the model, and conclude that the model is already optimized with respect to the considered refinements. We also compare it to two other computational stemness measures and conclude that the Ising model shows comparable performance. With further refinements, it will be a valuable computational measure for scoring the stemness of single cells, for simulating data from a population of cells, and for investigating the effect of a drug treatment or gene knockout. This will help the understanding of complex glioblastoma tumor tissue and will allow the exploration of possible treatments.