Browsing by Author "Grönfors, Markus"

In this thesis the main idea is analyse bacterial data obtained with specific technology called phenotype microarrays. The goal is to implement statistical methods and model cell respiration over period of 48 hours. The data are a bacterium called yersinia enterocolitica, which is a pathogen mainly carried by animals. Data was originally published in a scientific journal called Proceedings of the National Acedemy of Sciences of the United States of America and a small part of strain was chosen for thesis. Data consists about 110 000 rows of observations and it is divided to two experimental setups that are tested in two different temperatures. Data analysis consists three steps: cluster analysis, data normalization and comparing experimental setups. Statistical methods used are k-means clustering, Michaelis-Menten kinetics for growth curves, linear mixed effects models, restricted maximum likelihood estimation, random walk Metropolis-Hastings algorithm and highest posterior density intervals. Main results are there is a recognizable cluster for substrates implying grow and there are no differences between experimental setups. In conclusion statistical methods used in thesis are satisfactory for modelling data and while there are noticeable clusters, there lies no differences between experimental setups. In further analyses it should be better to include more experimental setups in one analysis.