Results Table HELP
Consecutive number of the prediction in the table ordered by the selected
column. To sort results by the selected column click on the small triangle
next to the column title. Results in the sorted column are bolded.
Rank of the prediction among all predictions submitted on the selected
target according to the GDT_TS score.
Name of the model as assigned at the time of prediction acceptance.
Name of each model is composed from the number of the target,
prediction format category, prediction group number, and model index.
If a model has more than one independent segment they are in separate files
with additional index indicating their sequential position within the
model.
Letter 'u' appended to the index of model or independent
segment indicates UNREFINED set of coordinates.
Example:
Model name T0201TS012_2u has the following components:
T0201 target number
TS Tertiary Structure (3D atoms coordinates) prediction
012 prediction group 12
2u model index 2, UNREFINED set of coordinates
Number of the predicting group received at the registration. Letter "s" placed
next to the group number identifies the group as taking part in the server-CASP
experiment.
Name of the predicting group.
|
A
|
Flat text file (output from
ACE
program) and LGA which contains
most of the numerical data available through this browser.
|
|
D
|
Target and model structures superimposed in sequence dependent manner.
Only CA atom pairs closer than 4.0 Å were used to obtain this
superposition. The GDT procedure was used to select the set of CA atom pairs.
Coloring scheme is based on distances between corresponding residues:
|Ti-Mi|< 1 Å: CYAN
|Ti-Mi|< 2 Å: GREEN
|Ti-Mi| < 4 Å: YELLOW
|Ti-Mi| < 8 Å: ORANGE
|Ti-Mi| > 8 Å: PURPLE for target, RED for model
|
|
I
|
Target and model structures superimposed in sequence independent manner.
This superposition was generated using
LGA
structural alignment, i.e.
completely ignoring sequence relationship between target and model.
Coloring scheme is based on alignment accuracy for corresponding residues:
perfectly aligned: GREEN
within 4 residues: YELLOW
not aligned: PURPLE for target, RED for model
Note: Structural superposition is not shown when
model had less than 10 residues.
|
|
G
|
Plot shows Target-Model
GDT (Global Distance Test) results.
|
GDT_TS - GlobalDistanceTest_TotalScore
GDT_TS = (GDT_P1 + GDT_P2 + GDT_P4 + GDT_P8)/4,
where GDT_Pn denotes percent of residues under distance cutoff <= nÅ
GDT_HA - GDT High Accuracy
GDT_HA = (GDT_P0.5 + GDT_P1 + GDT_P2 + GDT_P4)/4,
where GDT_Pn denotes percent of residues under distance cutoff <= nÅ
GDC_SC - Global Distance Calculation for sidechains
GDC_SC is GDT-like metrics wich uses a characteristic atom near the end of
each sidechain type (instead of CA's) for the evaluation of residue-residue
distance deviations.
GDC_SC = 2*(k*GDC_P1 + (k-1)*GDC_P2 ... + 1*GDC_Pk)/(k+1)*k, k=10
where GDC_Pk denotes percent of residues under distance cutoff <= 0.5kÅ
GDC_ALL - Global Distance Calculation for all atoms
GDC_ALL = 2*(k*GDC_P1 + (k-1)*GDC_P2 ... + 1*GDC_Pk)/(k+1)*k, k=10
where GDC_Pk denotes percent of atoms under distance cutoff <= 0.5kÅ
Z-M1-GDT (applicable to models No.1 only) - z-score showing relative quality
of the model among the first models submitted on the target by both human and server
groups (based on the GDT_TS score).
Z-M1s-GDT (applicable to models No.1 and server groups only) - z-score showing relative quality
of the model among the first models submitted on the target by server groups (based on the GDT_TS score).
Z-MA-GDT - z-score showing relative quality of the model among all models submitted on the target
by both human and server groups (based on the GDT_TS score).
Z-MAs-GDT (applicable to server groups only) - z-score showing relative quality of the model among
all models submitted on the target by server groups (based on the GDT_TS score).
RMS_CA - Root-mean-square deviation for the entire target structure or a subdomain, calculated on CA atoms and based on a sequence-dependent superposition.
RMS_ALL - Root-mean-square deviation for the entire target structure or a subdomain, calculated on all atoms and based on a sequence-dependent superposition.
RMSD - Root-mean-square deviation for the subset of CA atoms from the model that correspond to the residues from target structure or a subdomain in the sequence-independent LGA superposition.
Number of residues in a prediction submitted on the target or a subdomain.
This is a subset of residues in the prediction that overlap by sequence with the residues
possible to evaluate for that target/domain.
Number of residues in target (as received from experimentalists) or a subdomain
(as defined by assessors). In other words, number of residues possible to evaluate.
Number of residues in model.
NP_P = NP / NT * 100% - percent of residues predicted in the model from among those possible to evaluate.
Percent of non-hydrogen atoms predicted in the model from among those possible to evaluate.
Total number of CA-CA irregularities in the prediction.
Irregularity is reported if
0.1Å<|CAi-CAj|<3.6Å or
|CAi-CAi-1|>4.0Å.
Very closely located CA atoms (|CAi-CAj|<0.1Å)
are not included into the total number of irregularities.
Detailed analysis of CA-CA irregularities can be found on the GDT pages.
Mean distance between consecutive CA atoms in the model.
Standard deviation of distances between consecutive CA atoms in the model.
This column contains results of rmsd-based clustering of predictions and reports number of predictions that are similar to the model (RMSD < 0.1Å).
Data numbers are links that lead to the model names that are similar to the current prediction.
Format of the record:
T0199TS033_4: 2: 164_1 352_4
| | -----------
Model | |____ Shortened names of similar predictions.
Number of 164_1 corresponds to T0199TS164_1 or T0199AL164_1
similar predictions
Number of other predictions that are similar to the model (RMSD < 0.5Å).
Data numbers are links that lead to the model names that are similar to the current prediction. Format of the record is the same as for PC01 (see above).
AL0_P = AL0 / NT * 100% - percent of residues correctly aligned in the model (from among those possible to evaluate) based on the LGA sequence independent superposition generated with a 4Å
distance cutoff.
AL0: Number of correctly aligned residues in the 4Å LGA superposition of the model and experimental structure of the target. A model residue is considered to be correctly aligned if the CA atom falls within 3.8Å of the corresponding experimental atom, and there is no other experimental structure CA atom nearer.
AL4_P = AL4 / NT * 100% - percent of residues that can be correctly aligned with allowance for 1-4 -residue shift based on the LGA sequence independent superposition generated with a 4Å distance cutoff.
AL4: Number of residues in the model for which the closest experimental structure CA atom is within +/-4 residues of the correct one, and the distance between the corresponding residues is less than 3.8Å (based on the LGA sequence independent superposition generated with a 4Å distance cutoff).
ALI_P = ALI / NT * 100% - percent of residues that can be correctly structurally aligned with allowance for more than 4-residue shift based on the LGA sequence independent superposition generated with a 4Å distance cutoff.
Z-M1-GDT (applicable to models No.1 only) - z-score showing relative quality
of the model among the first models submitted on the target by both human and server
groups (based on the AL0_P score).
Z-M1s-GDT (applicable to models No.1 and server groups only) - z-score showing relative quality
of the model among the first models submitted on the target by server groups (based on the AL0_P score).
Z-MA-GDT - z-score showing relative quality of the model among all models submitted on the target
by both human and server groups (based on the AL0_P score).
Z-MAs-GDT (applicable to server groups only) - z-score showing relative quality of the model among
all models submitted on the target by server groups (based on the AL0_P score).
ALI: Number of residues in the model for which there exists an aligned residue in the target structure located less than 3.8Å apart (based on the LGA sequence independent superposition generated with a 4Å distance cutoff).
EQV5_0 - percent of residues correctly aligned in the model (from among those NT possible to evaluate) based on the sequence independent superposition generated with a 4Å distance cutoff. A residue is assumed to be correctly aligned if a corresponding experimental structure residue in the LGA alignment is also the correct one on the sequence, and the distance between them is less than 4Å .
EQV5_4 = EQ4 / NT * 100% - percent of residues that can be correctly aligned with allowance for 1-4 -residue shift based on the sequence independent superposition generated with a 4Å distance cutoff.
EQ4: Number of residues in the model for which corresponding residue in the LGA alignment output is within +/-4 residues of the correct one, and the distance between the corresponding residues is less than 4Å .
EQV5_I = EQI / NT * 100% - percent of residues that can be correctly structurally aligned with allowance for any residue shift based on sequence independent superposition generated with a 4Å distance cutoff.
EQI: Number of residues in the model for which there exists a corresponding residue in the LGA alignment output and the distance between the corresponding residues is less than 4Å .
ProSA uses knowledge-based potentials of CA atoms. (ref.: ProSA web:...)
The Z-Score indicates overall model quality and measures the deviation of the total energy of the structure with respect to an energy distribution derived from random conformations.
Aggregated Molprobity Score calculated according to the formula:
MP-Score = 0.426 *ln(1 + Clash-Score) + 0.33 *ln(1 + max(0, Rot-out - 1)) + 0.25 *ln( 1 + max(0, (100 - Ram-fv) - 2 )) + 0.5.
A structure with a MolProbity score of 3.0 or less is considered good.
Molprobity Clash Score - the number of all-atom steric overlaps > 0.4Å per 1000 atoms.
A clash score below 2.7 is considered very good, as 2.7 is the average clash score observed in the best parts of high-resolution structures.
Molprobity Rotamer Outliers Score - percentage of sidechains conformations classified as rotamer outliers, from those sidechains that can be evaluated.
Molprobity Ramachandran Outliers Score - percentage of backbone Ramachandran conformations classified as outliers.
Molprobity Ramachandran Favored Score - percentage of backbone Ramachandran conformations in favored region.
The average value of Sphere Grinder Score calculated for two sets of parameters: spheres' radii = 6Å in both cases and RMSD-cutoffs = 2Å and 4Å.
For every residue RMSD between the model and the target is calculated on the set of atoms inside the sphere of a certain radius with the center at CA.
The Sphere Grinder Score (SG) - percentage of residues with RMSD under the cutoff.
The lDDT score is computed by comparing inter-atomic interactions in models and targets.
(ref.:CASP9 TBM Assessment)
CAD-score evaluates protein models against the target stucture by quantifying differences between contact areas.
(ref.: CAD-score:...)
AA - all atoms are calculated;
SS - atoms of sidechains are taken into account only
Distances between all N/C atoms within Dmax are computed for both the
Target and Model structures. Atom pairs within Dmax in both structures
are counted as TP. Atom pairs which is less than Dmax in the Target
structure, but greater than Dmax in the Model structures are counted
as FN. Atom pairs which is greater than Dmax in the Target structure,
but less than Dmax in the Model structure are counted as FP. Recall,
Precision and F2-measure are calculated accordingly. A F2-measure
score for a random structure is also calculated. The RPF score is a
normalized F2-measure score where RPF(random) = 0 and RPF(Target) = 1.
Dmax = 9Å.
Quality Control Score (QCS) is a method to evaluate the quality of
structure predictions, especially for template free structure prediction.
It measures and scores correctness of secondary structure elements (SSE) predictions,
their relative positions, handedness of SSE triplets, all C-alpha contacts.
(ref.: An automatic method ...)
Contact score based on all C-alpha from QCS bundle.
SOV score measures the correctness of secondary structure predictions.
The CASP TS predictions are translated into SS predictions by dssp.
(ref:sov_help)
TMscore measures the similarity of two protein structures.
(ref:Scoring function for ...)
CE (Combinatorial Extension) is a method for comparing and aligning protein structures.
(ref:Protein structure alignment ...)
DFM (deformation score) is calculated as a measure of distortion of correspondinf tetrahedrons
with vertices at atoms at target and model structures.
(ref:Assessment of template-free ...)
Handedness score is calculated as a fraction of the same-way oriented tetrahedrons with vertices at C-alpha's
at target and model structures in the procedure of randomly selecting atoms and repeating 50,000 times.
(ref:Assessment of template-free ...)
CoDM score is a weighted Pearson's correlation of the distance matrices of the target and model structures.
(ref:Assessment of template-free ...)
The score is based on the energy of elastic deformation between two structures: model and target.
(ref:FlexE: Using Elasic Network Models to ...)
ASE (Accuracy Self Estimate) score is based on the accuracy of every residue position in the model reported in the temperature factor (B-factor) field.
Up to CASP15, structure predictors were asked to submit per-residue error estimates in Angstroms as a measure of local accuracy of own models. Correspondingly, the ASE score was calculated by formula:
ASE = 100.0*(1 - Mean(|S(tfi/d0)-S(di/d0)|))
where tfi - temperature factor (accuracy self-estimate) of the i-th residue in the model,
di - distance between the i-th residues in a model and the target in the lga sequence-dependent superposition,
S(x) = 1/(1+x^2), and d0 - scaling factor set to d0=5.0.
Starting from CASP15, the predicted LDDT score (pLDDT) is requested as a measure of local accuracy self-estimation. Correspondingly, the formula for ASE calculation changed to
ASE = 100 - Mean(|pLDDTi - LDDTi|)),
where the pLDDT and LDDT scores are both in 0-100 range.
reLLG score of 10 or higher indicate models that most likely will be successful for molecular replacement.
The relative eLLG score is an approximation of how well the model would behave in molecular replacement, compared to the quality of model that would come from another crystal form of the same protein. The score gives the relative log-likelihood-gain (LLG) for the model being evaluated, as a percentage of the LLG expected from a model with about 0.4 Å RMS error from the target structure. The difference between the B_as_pLDDT and B_constant scores indicates how much value has been added by the model confidence measure.
Only very good models do well by this score, and models with substantial errors (such as incorrect relative domain orientations) are heavily penalized, as is the case when carrying out molecular replacement against experimental diffraction data.
(ref:Assessing the utility of CASP14 models ...)
DipDiff is the average difference between the local DipScores of the target and a given model.
DipScore measures the likelihood of the backbone distances around a given CA to be correct.
(ref:High accuracy protein structure prediction ...)
SC dihedral error score is a measure of side chain quality based on the divergence of observed model dihedral angles chi1 and chi2 vs the target: chi1 is given a greater weight than chi2 in the scoring and the degree of burial of the residue is also taken into account. The score ranges from 0 (best) to 1 (worst).
(ref:Evaluation of template-based modeling ...)
BB dihedral error score is a measure of local backbone quality based on the divergence of observed model dihedral angles phi, psi and omega (equally weighted) vs the target. The score ranges from 0 (best) to 1 (worst).
(ref:Evaluation of template-based modeling ...)
Column "Pclust" contains 2 values showing results of rmsd-based clustering of predictions. First column (0.1) reports number of predictions that are very similar to each other (RMSD < 0.1Å). Second column (0.5) shows number of predictions that are similar within the range of 0.5Å. In both cases links lead to the model names that are similar to the current prediction.
Format of the record:
T0199TS033_4: 2: 164_1 352_4
| | -----------
Model | |____ Shortened names of similar predictions.
Number of 164_1 corresponds to T0199TS164_1 or T0199AL164_1
similar predictions
Format of the column:
10/3-40(2)
| | | |
| | | C4: Number of clusters of identical CA atoms
C1: Number of CA-CA irregularities | |
| C3: Total number of identical CA atoms
C2: Number of severe collisions
C1 (0.1Å<|CAi-CAj|<3.6Å or
<|CAi-CAi-1|>4.0Å)
CA-CA irregularity is reported if
-
distance between any two CA atoms in the prediction is smaller than 3.6Å
-
or distance between the two consecutive CA atoms is bigger than 4.0Å.
Very closely located CA atoms (|CAi-CAj|<0.1Å)
are not included into the total number of irregularities C1 and are
counted separately (C3).
C2 (0.1Å<|CAi-CAj|<1.9Å)
Severe collisions are reported if distance between any two CA atoms is less
then twice as short as the regular distance between two consecutive CAs.
Very closely located CA atoms (|CAi-CAj|<0.1Å)
are not included into the total number of severe collisions C2 and are
counted separately (C3).
C3 (|CAi-CAj|<0.1Å)
C3 represents the total number of very closely located CA atoms in the prediction.
These atoms may belong to different clusters of identical atoms (C4).
For example, the prediction that contains 3 identical CA atoms
ATOM 441 CA SER 51 0.000 0.000 0.000 1.00 0.00
ATOM 446 CA ALA 52 0.000 0.000 0.000 1.00 0.00
ATOM 451 CA SER 53 0.000 0.000 0.000 1.00 0.00
and 2 CA atoms with identical coordinates but different from the first group
ATOM 1441 CA SER 251 1.000 2.000 3.000 1.00 0.00
ATOM 1446 CA ALA 252 1.000 2.000 3.000 1.00 0.00
will be assigned number 5 (3+2) as C3 and 2 (different clusters) as C4.
C4
C4 shows how many different clusters very closely located CA atoms
belong to (see explanation for C3).
-------------------------------------
The hyperlink for non-zero entries in the "CA-CA_irreg" column points
to the detailed information about CA-CA clashes in the prediction.
Format:
50 51 13.40
#res1 #res2 distance(Å)