SCITE method can reconstruct the course of development of cancer in cells from data on their mutations. The course of development is represented with a mutation tree, which is similar to a phylogenetic tree. In the thesis, we implement some functionalities of the SCITE method with the Python programming language and ensure comparable execution time. We focus on the posterior distributions of mutation trees and the probabilities of overlooked mutations in cells. The method is improved with the introduction of partial mutation tree scoring which speeds up the scoring process. Along the existing tree moves we propose a new tree move and prove its usefulness on multiple datasets. Lastly, we enable the user to compute the effective sample size of posterior samples and thus enable better assessment of the results of the algorithm.
|
