Alternative views

Results page provides viewing of the results target-by-target. Groups view allows assessing performance of a particular prediction group over all the targets predicted by that group. Both the re-compiled tables and graphical results - with predictions from the group in question specially indicated - are available. Table Browser adds the capability to generate custom comparisons of numerical results where prediction groups, targets, and measures may be selected by the user.

Results page

The Results page allows access to (1) tabulated numerical results (link on top of the target panel), (2) GDT plots, (3) alignment quality bar graphs. Targets may be selected based on prediction category (CM ? comparative modeling, FR ? fold recognition, NF ? new fold, FR targets may further be subdivided into FR/Analogous and FR/Homologous. See also Categories of prediction).

(2) Tables

Models are identified with target and prediction group numbers. Group name and results of the sequence-dependent (SD) and sequence-independent (SI) analyses, as well as DAL local structure analysis are provided for each model. Text files, interactive renderings of model and target in a sequence dependent and sequence independent superpositions, as well as links to the GDT plots are also provided (links through letters A, D, I, and G respectively). Tables may be sorted according to results in each column and may be expanded or contracted with full/brief options. Additional descriptions (e.g. of measures) are provided through links in the tables.

(3) GDT plots

(3) GDT plots These plots provide analysis of prediction quality by finding the largest subset of residues in the model that can be fitted to the target in a rigid body sequence dependent superposition. This is done for cutoffs ranging from 0.5 to 10.0 Angstroms. Results are plotted as a line for each model separately. Models from several groups may be displayed simultaneously; and conversely, clicking on any line in the plot will identify the corresponding model and research group (dark blue indicates Model 1, cyan - other models submitted by that group).

(4) Alignment quality

The bar graphs indicate percentages of correctly aligned residues in the model relative to target (green), those aligned with an error not larger than 4 residues (yellow), and those aligned with an error of 5 or larger (red). Additional set of bar graphs indicates the distribution of the correctly and incorrectly aligned residues along the target protein sequence. Clicking on each bar produces a 3D interactive rendering of a superimposed model and target in the best rigid body sequence independent superposition (obtained with LGA). Further details and definitions are provided through links in graphs.