Improved greedy algorithm for protein structure reconstruction |
| |
Authors: | Tuffery Pierre Guyon Frédéric Derreumaux Philippe |
| |
Institution: | Equipe de Bioinformatique Génomique et Moléculaire, INSERM E0346, Université Paris 7, Tour 53-54, 2 place Jussieu, 75251 Paris Cedex 05, France. tuffery@ebgm.jussieu.fr |
| |
Abstract: | This article concerns the development of an improved greedy algorithm for protein structure reconstruction. Our stochastic greedy algorithm, which attempts to locate the ground state of an approximate energy function, exploits the fact that protein structures consist of overlapping structural building blocks that are not independent. Application of this approach to a series of 16 proteins with 50-250 amino acids leads to predicted models deviating from the experimental structures by 0.5 A RMSD using an RMSD-based energy function and within 1.5 to 4.8 A RMSD using a Go-based energy function. The Go-based results are significant because they illustrate the strength of combining structural fragments and stochastic greedy algorithms in capturing the native structures of proteins stabilized by long-range interactions separated by more than 30 amino acids. These results clearly open the door to less computationally demanding solutions to predict structures from sequences. |
| |
Keywords: | protein structure reconstruction greedy algorithms discrete conformation space Go energy model structural alphabet |
本文献已被 PubMed 等数据库收录! |
|