ADVANCED COMPUTATIONAL METHODS FOR THE INVESTIGATION OF BIOPHYSICAL PROPERTIES OF MACROMOLECULAR PROTEINS

The focus of this thesis is the application of novel advanced computational methods to specifically interpret data from biophysical experiments performed on particular macromolecular structures: microtubules and tubulin. Microtubules (MTs) are macromolecular protein assemblies with well-known key roles in all eukaryotic cells. It is assumed however that they also have an important role in intercellular communication over long distances, especially in the network of neurons. This feature may explain mechanisms little-known so far, such as the immediate involvement of the entire immune system to local damage, or may help to clarify some unsolved questions about the mind-body problem.Taking into account the connection between structural and physical properties in Carbon Nanotubes (CNTs) and their structural similarity to MTs, our basic assumption in this research was that when tubulin and MTs show different biophysical behaviours, this should be due to the peculiar dynamical organization of MTs. In order to investigate the biophysical properties of the macromolecular structure object of this study, microtubules and tubulin, an innovative approach has been applied: - ad hoc experimental procedures have been prepared, from which the experimental data have been obtained; - ad hoc computational methods have been developed specifically to interpret the experimental data. The theoretical assumptions of the computational methods used are based on the theory of dynamic evolution of complex systems and on the properties originating from their self-organization capacity. The physical properties of birefringence, resonance and superradiance were measured, and data from biophysical experiments were analysed using ad hoc computational methods: dynamic simulation of MTs and tubulin was performed, and their level of self-organization was evaluated using artificial neural networks. The results from dynamic simulations were submitted to two different self-organizing artificial neural networks: the first one for the evaluation of specific parameters, and the second one for the evaluation of dynamic attractors. We developed a procedure that processes in the form of attractors the series of winner neurons resulting from the output of the neural network. Both tubulin and MTs show dynamic stability, but only MTs exhibit a significant behaviour in presence of electric field, in the direction of a stronger structural and spatial organization. The Artificial Intelligence approach supports the experimental evidences at the microscopic level, allowing a more correct and accurate interpretation of the results. The research we carried out reveals the existence of a dynamic and organized response of MT to electromagnetic fields, which could justify their role in receiving and transmitting information even at long distances. The innovative computational methods implemented in this work revealed to be very useful for the dynamic analysis of such complex structures.