Figure 3

Model quality improvement achieved by the use of multiple templates (relative to the single-template method, measured by GDT-TS) against average template quality improvement (relative to that of top ranker, measured by TM-score) for the 43 targets for which multiple templates are used. Only core residues were considered in this comparison. Overall, as the average quality of multiple templates decreased, model quality improvement achieved by the use of multiple templates tended to decrease for both model-building methods, MODELLER (empty dots) and MODELLER-CSA (filled dots). Targets for which the top ranking template was the best-quality template are shown in green and others in blue.