From Protein Structure to Function with Bioinformatics.pdf

3 Comparative Protein Structure Modelling 65the modelling procedure can construct a homology-based model of the whole targetsequence. Second, the template structures may be aligned with the same part of thetarget and build the model on the locally best template.An elaborate way to select suitable templates is to generate and evaluate models foreach candidate template structure and/or their combinations. The optimized all-atommodels can then be evaluated by an energy or scoring function, such as the Z-score ofPROSA (Sippl 1995) or VERIFY3D (Eisenberg et al. 1997). These scoring methods areoften sufficiently accurate to allow selection of the most accurate of the generated models(Wu et al. 2000). This trial-and-error approach can be viewed as limited threading (i.e.,the target sequence is threaded through similar template structures). However theseapproaches are good only at selecting various templates on a global level.A recently developed method M4T (Multiple Mapping Method with MultipleTemplates) selects and combines multiple template structures through an iterativeclustering approach that takes into account the “unique” contribution of each template,their sequence similarity among themselves and to the target sequence, andtheir experimental resolution (Fernandez-Fuentes et al. 2007a, b). The resulting modelssystematically outperformed models that were based on the single best template.Another important observation from this study was that below 40% sequenceidentity, models built using multiple templates are more accurate than those builtusing a single template only and this trend is accentuated as one moves into moreremote target-template pair cases. Meanwhile the advantage of using multiple templatesgradually disappears above 40% target-template sequence identity cases (Fig.3.1). This suggests that in this range the average differences between the template andtarget structures are smaller than the average differences among alternative templatestructures that are all highly similar to the target (Fernandez-Fuentes et al. 2007b).3.2.3 Sequence to Structure AlignmentTo build a model, all comparative modelling programs depend on a list of assumedstructural equivalences between the target and template residues. This list is definedby the alignment of the target and template sequences. Many template search methodswill produce such an alignment and these sometimes can directly be used as theinput for modelling. Often, however, especially in the difficult cases, this initialalignment is not the optimal target-template alignment e.g., at less than 30%sequence identity (where sequence identity is defined as the number of identicalpositions in the alignment normalized by the length of the target sequence). Searchmethods tend to be tuned for detection of remote relationships, which is often realizedbased on a local motif and not for a full length, optimal alignment. Therefore,once the templates are selected, an alignment method should be used to align themwith the target sequence. The alignment is relatively simple to obtain when the targettemplatesequence identity is above 40%. If the target-template sequence identity islower than 40%, the alignment accuracy becomes the most important factor affectingthe quality of the resulting model. A misalignment by only one residue position willresult in an error of approximately 4 Å in the model.