12.07.2015 Views

From Protein Structure to Function with Bioinformatics.pdf

From Protein Structure to Function with Bioinformatics.pdf

From Protein Structure to Function with Bioinformatics.pdf

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

3 Comparative <strong>Protein</strong> <strong>Structure</strong> Modelling 69etry approach constructs all-a<strong>to</strong>m models from lower and upper bounds on distancesand dihedral angles (Havel and Snow 1991). Although further efforts were made <strong>to</strong>apply distance geometry for comparative modelling, e.g. (Aszodi and Taylor 1996),more successful but also more conservative, real space modelling approaches dominatethe field, perhaps because evolution also proved <strong>to</strong> be surprisingly conservativein preserving structural features in various proteins (Kihara and Skolnick 2003).Comparative modelling by satisfaction of spatial restraints is implemented inthe computer program MODELLER (Fiser and Sali 2003a; Sali and Blundell1993), currently the most popular protein modelling program. In the first step ofmodel building, distance and dihedral angle restraints on the target sequence arederived from its alignment <strong>with</strong> template 3D structures. The form of theserestraints was obtained from a statistical analysis of the relationships betweensimilar protein structures. By scanning the database of alignments, tables quantifyingvarious correlations were obtained, such as the correlations between twoequivalent Cα-Cα distances, or between equivalent main chain dihedral anglesfrom two related proteins (Sali and Blundell 1993). These relationships areexpressed as conditional probability density functions (<strong>pdf</strong>’s) and can be useddirectly as spatial restraints. For example, probabilities for different values of themain chain dihedral angles are calculated from the type of residue considered,from the main chain conformation of an equivalent template residue, and fromsequence similarity between the two proteins. An important feature of the methodis that the forms of spatial restraints were obtained empirically, from a database ofprotein structure alignments, <strong>with</strong>out any user imposed subjective assumption.Finally, the model is obtained by optimizing the objective function in Cartesianspace. The optimization is carried out by the use of the variable target functionmethod (Braun and Go 1985) employing methods of conjugate gradients andmolecular dynamics <strong>with</strong> simulated annealing (Clore et al. 1986).A similar comprehensive package is NEST that can build a homology modelbased on single sequence-template alignment or from multiple templates. It can alsoconsider different structures for different parts of the target (Petrey et al. 2003).Combining Alignments, Combining <strong>Structure</strong>sIt is frequently difficult <strong>to</strong> select the best templates or calculate a good alignment.One way of improving a comparative model in such cases is <strong>to</strong> proceed <strong>with</strong> an iterationof template selection, alignment, and model building, guided by model assessment.This iteration can be repeated until no improvement in the model is detected(Fiser et al. 2003; Guenther et al. 1997). More recently these anecdotal and manualapproaches were au<strong>to</strong>mated (Petrey et al. 2003). For instance, an au<strong>to</strong>mated methodwas introduced that optimizes both the alignment and the model implied by it (Johnand Sali 2003). This task is achieved by a genetic algorithm pro<strong>to</strong>col that starts <strong>with</strong>a set of initial alignments and then iterates through re-alignment, model building andmodel assessment <strong>to</strong> optimize a model assessment score. During this iterative processnew alignments are constructed by application of a number of opera<strong>to</strong>rs, such asalignment mutations and cross-overs; comparative models corresponding <strong>to</strong> these

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!