DERIVATION OF A REWEIGHTING ALGORITHM FOR THE PREDICTION OF THE THERMODYNAMIC STABILITY OF MUTATED SEQUENCES AND EVALUATION OF ITS USABILITY IN A SIMPLE CONTINUOUS PROTEIN MODEL

Abstract: In this thesis, a new reweighting algorithm for predicting the native conformations of amino acid sequences, consisting of 16 residues in a reduced representation continuous protein model, is derived and tested to map neutral nets of α-helices and β-sheets. For examining the potential of the reweighting algorithm for predicting the thermodynamic stability of amino acid sequences, it is applied to a mutational pathway, connecting a α-helix to a β-sheet. Moreover, a bistable sequence is used to predict the evolution of the chain properties along the pathway. The reweighted averages of four relevant observables are then compared to the averages obtained in direct simulations. Further, neutral nets of α-helices and β-sheets are mapped by using the reweighting algorithm. The compositions of the neutral nets are analyzed and the validity of the mapping is tested, both, in direct simulations and the analysis of a few sequences. The evaluation of the usability of the reweighting algorithm indicates, that, in principle, it is capable of predicting the stability of native conformations. Moreover, the test cases show that for this method, bistable sequences are the most appropriate starting point for making predictions of the stability of other sequences. In the analysis of the neutral nets, it is shown that the compositions of the nets display some realistic features. However, as it turns out, some sequences, that are identified as belonging to either of the two nets do not fulfill the requirements that define these nets.