IFLP   13074
Unidad Ejecutora - UE
congresos y reuniones científicas
Protein Secondary structures determine critical size moves for protein folding in the Monte Carlo Simulations
Sao Paulo
Congreso; 3rd Conference of the Brazilian Association for Bioinformatics and Computational Biology; 2007
Institución organizadora:
Asociacion Brasilera de Bioinformatica y Biologia Computacional
There is an enormous amount of amino acid sequence data currently available from a great many genes from many organisms, and this database increases from day to day. However, our ability to fully exploit this data is limited by our inability to predict protein structure directly from its aminoacid sequence. The access to the folded structure of a protein is essential for a better understanding of its stereochemistry, which indirectly determines its biochemical function.It is well known that the difficulties in predicting protein structure lie in the combination of their very large conformational space and the rough nature of the energy landscape defined over this space. Consequently much effort has been dedicated to the creation of algorithms for an e¢ cient search in the protein conformational space. In general, methods capable of sampling the more stable states of the protein chain are needed for the determination of the native structure ofa protein. The secondary, tertiary structures of proteins are the main conformations to be sampled from an energetic point-of-view. However, a minimum energy criterion is notenough for the establishment of the native structure. An efficient method that searches over the stable states in a feasible computation time is also required. In the MC simulations of a simple on-lattice protein model the degree of freedom is reduced and it is possible to exhaust the phase space and to determine any observable of the system accurately. The usual chain moves in these models are sufficient to guarantee the ergodic hypothesis. On the other hand, one needs to be careful with MC studies performed with all-atoms models. This is because high amplitude moves are associated with low acceptance rate, while very small moves are inefficient to obtain a representative state sample, leading to a loss of ergodicity in both cases. In general,algorithms that do not obey these rules (such as the pivot algorithm) must include complementary methods so that they are successful. However, depending on the size of the system, these calculations can be unfeasible. An efficient searching method will use movements that allow access from any state to another in a feasible simulation time. In this sense, several approaches have been proposed to overcoming such computational difficulties so that the convergence ratecan be improved. Recently, a Monte Carlo based algorithm able to produce local configurations in all-atom protein models has been introduced. This algorithm, calledLMProt (Local Moves for Proteins), combines the ability to control the size of the movements r (defined as the maximum atom displacement in the movements) with the capacity to control the flexibility of the chain. LMProt uses local moves (LM) defined by the movement of a small segment of residues randomly chosen within the interval {4-6}, and it is capable of efficiently folding large proteins (>100 residues) using an artificial potential. In the present work, we explore the effect of size movements r on protein folding effiency in detail. Particularly, we study the effect of r during the formation of secondary and tertiary structures, emphasizing the in‡uence of stereochemical effects on the way the chain is moved by the MC methods. A number of methods have recently been proposed to produce local moves to sample protein conformational space more efficiently. Despite the critical role of size moves in simulated protein folding, there are no studies in this direction. We have identified that Monte Carlo based algorithms for protein folding by local moves, have a critical size move rc. In the folding simulations of 16 proteins using the LMProt algorithm, we found that size move smaller than the rc have a very poor folding success rate. Our results indicates that the rc-value obtained is associated with stereochemi-cal restrictions of secondary structures. We also report that the folding simulation of alpha-type protein is more sensitive to the size movement when compared to beta-type protein. The results reported in this paper provide a more precise view of local moves on protein folding simulationsand could be helpful in providing information to define a consistent Monte Carlo update.