Bayesian cluster analysis for single-molecule localization microscopy

Abstract

This thesis aims to develop and assess a software solution for Bayesian-based cluster analysis of super-resolution single-molecule localisation microscopy data and establish a methodology for imaging cells immersed in a liquid environment. Previous studies demonstrated that cluster analysis is susceptible to user bias, and advanced techniques require coding proficiency. Furthermore, previous research demonstrated the potential of DNA tethering for investigating clustering phenomena triggered by external stimuli without requiring cell adhesion. Nevertheless, DNA tethering adoption has been limited due to the requisite expertise in advanced chemistry. As a result, this dissertation focuses on developing a user-friendly software solution. The primary objectives are facilitating Bayesian-model-based cluster analysis and simplifying DNA-tethering. The designed DNA-tethering assay facilitates cell sample examination without surface adhesion. Consequently, researchers can investigate cell behaviour under specific and localised external stimuli in live-cell and single-molecule localisation microscopy modes. The software solution presented in this research builds upon previously published work, aiming to improve performance and accessibility. Additionally, it investigates the impact and limitations of different clustering algorithms on single-molecule fluorescence microscopy data. The software was implemented and evaluated using simulated and experimental data to validate its performance. Furthermore, the software could conduct cluster analysis on data acquired from the simplified DNA-tethering assay developed. This study presents a modular software tool with a user-friendly graphical user interface designed to facilitate cluster analysis of single-molecule localisation microscopy data. The software demonstrates improved computational and storage capabilities, empowering researchers to evaluate samples effectively. Furthermore, this software solution is suitable for all clustering-related single-molecule localisation microscopy research studies and can be customised with additional features.