Protein structure prediction and folding dynamics
The topic of protein folding can be studied from two different points of view. The first is concerned with the question of how the biologically relevant three-dimensional structure can be determined from a given amino acid sequence. This is of great practical interest as experimental determination o...
| Main Author: | |
|---|---|
| Format: | Others |
| Language: | English en |
| Published: |
2010
|
| Online Access: | http://tuprints.ulb.tu-darmstadt.de/2068/1/diss_kwolff_genehmigt.pdf Wolff, Katrin <http://tuprints.ulb.tu-darmstadt.de/view/person/Wolff=3AKatrin=3A=3A.html> : Protein structure prediction and folding dynamics. Technische Universität, Darmstadt [Ph.D. Thesis], (2010) |
| Summary: | The topic of protein folding can be studied from two different points of view. The first is concerned with the question of how the biologically relevant three-dimensional structure can be determined from a given amino acid sequence. This is of great practical interest as experimental determination of protein structure is difficult and costly, whereas sequencing is relatively simple and cheap. The second question is that of the physical process of folding where often knowledge of the biologically active (native) structure is assumed. As protein misfolding is the cause of several diseases, this is of bio-medical importance as well. Moreover, it is also of fundamental interest as a mesoscopic system displaying cooperative effects. This thesis covers these two aspects of protein folding. To this end so-called structural profiles are defined which may act as a link between protein sequence and structure for the task of structure prediction. On the other hand they contain structure information in a compressed form which, as will be shown, also encodes the folding process. A severe bottleneck in protein structure prediction is the transition from a coarse-grained to a more detailed structure description and the refinement of structure candidates that are already close to the target structure. As refinement is very computation-intensive it is advisable to concentrate on a selection of promising candidates. This is where structural profiles predicted from sequence prove to be advantageous. Their usefulness in filtering is at least on par to established methods and they are clearly superior if the structure set is of only moderate quality or if the criterion is especially strict on what is to be considered a good structure. An important question regarding the folding process is in how far the structure of the native state dictates the folding pathway, thus allowing to abstract from the chemical details of the amino acid sequence. One class of such native-centric models are so-called Go-models which additionally rely on the principle of minimum frustration, meaning that only those amino acids that are in contact in the native state attract each other. In contrast, the model presented here, which is based on structural profiles, allows non-native interactions. Applying adapted sampling schemes to both native-centric models and to three well-studied example proteins shows that experimental results and behaviour observed in detailed all-atom simulations can be better explained in the profile-based model than in the Go-model. In particular, the profile-based model shows a cooperative folding transition and existence of secondary structure in the unfolded state. Thus, while a simple model based on pairwise contacts cannot adequately describe folding behaviour, the profile-based model, which is native-centric as well, can reproduce fundamental experimental observations. |
|---|