|
|
Distribution of selected evaluation scores for different types of models
For each measure (specified in the x-axis title), a blue boxplot shows the score distribution for models built starting from reference structure, while a red boxplot - for models built ab initio. Box boundaries correspond to the Q1=25th (bottom) and Q3=75th (top) percentiles in the data; the vertical line inside the box corresponds to the median (Q2). The width of the box defines the interquartile range (IQR=Q3-Q1). The length of the whiskers shows the range of the values outside the interquartile range, but within 1.5 IQR. The dots correspond to outliers, i.e. values outside the 1.5 IQR range.
|
|
Self-estimates (coords only)
![Cumulative MolProbity Score (>0):
Molprb = 0.426*ln(1+Clash) + 0.33*ln(1+max(0,Rot_out-1)) + 0.25*ln(1+max(0,(100-Ram_fv)-2)) + 0.5.
Scores below 3 usually indicate models of acceptable polypeptide geometry.
[smaller is better]](images/boxplots/molprb_mon.png)
![MolProbity Clash Score (>0):
the number of all-atom steric overlaps (>0.4Å) per 1000 atoms.
A good quality structure usually has a Clash-score below 20.
[smaller is better]](images/boxplots/mp_clash.png)
![ProQ global score:
model accuracy estimate (0-1) from the ProQ3 method, which combines knowledge-based energy terms and comparison of predicted and observed structural features. The score estimates overall reliability of the model.
[larger is better]](images/boxplots/proq3.png)
|
Fit to Map
![CCC - Cross-Correlation Coefficient:
the TemPy CCC score (0-1) is calculated by the array multiplication of density values at the same points in the model and target maps.
[larger is better]](images/boxplots/ccc.png)
![LAP - LAPlacian-filtered cross-correlation coefficient:
the TemPy LAP score (0-1) is computed similarly to TemPy's CCC score (above), using density maps pre-processed with a Laplacian filter.
[larger is better]](images/boxplots/lap.png)
![Overall FSC (Fourier Shell Correlation):
the PHENIX overall Fourier Shell Correlation (0-1) is calculated between the complex-valued Fourier map and model coefficients binned in resolution shells.
[larger is better]](images/boxplots/fsc.png)
|
Fit to reference structure
![RMSD:
A Root-Mean-Square Deviation calculated on Cα atoms of the representative subunits.
[smaller is better]](images/boxplots/rmsd.png)
![GDT_TS - GlobalDistanceTest_TotalScore:
a sequence-dependent score (0-100) reporting the average percentage of model Cα atoms that can be superimposed with the target structure under 1, 2, 4, and 8Å distance cutoffs.
GDT_TS scores over 50 indicate structures with significant similarity, while scores below 25 indicate unrelated structures (poor models).
[larger is better]](images/boxplots/gdt_ts.png)
![CAD-score:
a superposition-free measure (0-1) that estimates similarity of two structures by quantifying differences in their interatom contact areas.
[larger is better]](images/boxplots/cad.png)
![The QS score quantifies the similarity between quaternary structures in terms of shared interfacial contacts of their subunits.
The QS_global score reports fraction of shared contacts for all fitting interfaces.
[larger is better]](images/boxplots/qs_global.png)
|
Similarity to other models
![DAVIS_QA score:
consensus-based model accuracy estimate (0-1), which reflects similarity of the model to all other models submitted on the target.
[larger is better]](images/boxplots/davis_qa.png)
|
|
NOTE: Clicking on a set of boxplots brings up a new window with distribution plots for all measures in the selected category (e.g., Fit to Map).
|