The aim of this project is absolute quantification of heteroplasmy in mtDNA, motivated by the study of health issues related to changes in mtDNA structure. An initial examination of scRNAseq data in relation to the project aim is given, with an ideal scRNAseq pipeline discussed, with discrepancies outlined. The goal subsequently is to perform absolute inference on qPCR data to obtain values for initial copy number of mtDNA in single cells, amongst other target parameters. The binomial branching process Hidden Markov model from Lalam, 2007 is accepted as the model describing the process. Bayesian inference is implemented, with Metropolis Hastings as the MCMC method for obtaining the posterior. Additionally, Adaptive Metropolis is also implemented, for comparison and investigation into how well posteriors are inferred. Extensive results for variations of implementations are shown and discussed. The discussion scrutinizes results obtained, and considers possible adjustments for improving performance. Avenues to explore for future development include extending execution times, experimenting with priors, and applying analogous analysis to ddPCR data.
