Vorschau

PDF, Englisch Download (49MB) | Nutzungsbedingungen

Abstract

It is well known that our genetic material influences our tendency to develop certain conditions. Finding the causes behind these predispositions assumes the understanding of mechanisms handling and maintaining the genome. While the problem is important from the biological point of view, being one of the basic riddles of life, it also poses interesting questions which may only be answered by physics. Topics include transport, reaction-diffusion, polymer physics, equilibrium and non-equilibrium dynamics and chaos, amongst others. Experimental techniques, like microscopy or molecular biology approaches provide an ever improving insight in the structure of the nucleus, however, computational and modelling approaches are still needed to explain unknown aspects of genetics.

% Evidence is accumulating that our genetic material not only influences our resemblance to relatives and the chances that we may have a tendency to develop certain diseases, but also our predisposition to contract viral infections or to develop conditions like depression, obesity or substance dependence. It has become clear that understanding how the genetic material is organized and how it is being handled might be the key to revolutionize medicine. Experimental techniques, like microscopy or molecular biology approaches provide an ever improving insight in the structure of the nucleus, however, computational and modelling approaches are still needed to explain unknown aspects of genetics.

In this thesis we tackle the problem of understanding the structure of the nucleus from the two opposite sides of the experimental ``blind-spot''. We develop alternative image modelling and analysis tools which are able to capture and recreate the ``large scale'' density patterns observed in confocal microscopy images of the nucleus. For this, we introduce a generalized Potts model which is extensively analysed also from the statistical mechanics point of view. Furthermore, we apply statistical mechanics and graph theory calculations to study patterns registered with super resolution microscopy techniques. We investigate the effect of irradiation and light stress on the structure of the chromatin, and are able to quantitatively support prior experimental observations regarding structural changes.

Understanding the interaction and classification of proteins, structures which perform vastly different functions on molecular scales, is also important to achieve the final picture. We contribute to this by elaborating a framework to assess topological similarity among these chemicals. Our approach is based on recently developed computational topology algorithms used to calculate fingerprints of the molecules. We discuss three different modifications of the framework and investigate them on real-world datasets. In addition, we recognize that the mentioned fingerprints can be used to calculate the fractal dimension of certain objects, and offer an intuitive explanation for the observed relation.