Second Meeting on the Critical Assessment of
Techniques for Protein Structure Prediction
From this pop-up menu the id of the prediction target is selected. Target descriptions can be found in the Table of Fold-Recognition Targets at the bottom of the Fold Recognition Evaluation page.
Structure comparison method
For CASP2, predictions in the category threading/fold-recognition were evaluated
by comparing them to results from structure-comparison searches as
standards of truth. For most predictions, results evaluated against three different
structure-comparison methods are available. The structure comparison methods
used were
VAST,
DALI,
and SSAP.
Prediction team
32 different teams have participated in the category threading/fold-recognition
at CASP2. Select from the following:
Level of detail
This pop-up menu has two settings: "condensed" and "full-length". If the full-length
form is selected, the tables contain all the quantities that have been calculated
and are available. The condensed form produces smaller tables (which are faster
do download too!) with a few selected quantities only. For the Threading result
table these are:
Sort variable
A large number of different evaluation quantities have been calculated, and the
tables can be sorted by all of them, even if they are not displayed in the
"condensed" mode. This pop-up menu selects the evaluation
quantity the sort is based on, and quantities fall into three classes:
Sort order
This menu selects whether table rows are sorted in descending or ascending order
of the selected sort variable. Descending order means that
"better" predictions appear at the top of the table for most Alignment-quality quantities,
except for theAlignment RMS-deviation[ARms] and the
Alignment mean shift error[Shft].
In CASP2, each prediction target had been assigned a unique target identification
number, T00xx. This number is listed in the first column and it is used to
select targets in the results browser. In this table, target id numbers are linked
to the
original target submissions, as sent in by the crystallographers/NMR-spectroscopists.
Target nickname [Name]:
These are short abbreviations of the target protein names, used to refer to the target
sequences in CASP2. Nicknames are linked to
Entrez-reports on the respective proteins. Reports were selected by the amount of
annotation they contain.
Target size [Nres]:
This column gives the number of residues for each target sequence. Residue numbers are
not crosslinked to anything.
CASP2 prediction category [Category]:
In CASP2, predictions could be submitted in 4 different categories. Here, only targets
that fell into the fold-recognition/threading category are listed. Most of them have been
targets for ab-initio predictions as well, and some of them have been targets in the
comparativ modeling category too.
Structure determination method [Method]:
Structures of CASP2 targets have been determined by X-ray diffraction on protein crystals
or by NMR-spectroscopy of proteins in solution.
Protein Data-Bank accession [PDB-Id]:
Upon entry into the Protein Data Bank (PDB), each
structure is assigned a unique identifier to serve as an accession code. These codes
are listed for the CASP2 target proteins, and they are linked to three-dimensional
coordinates, if available from
MMDB or the PDB.
Structure determination status [solved]:
This column indicates whether the structure of the target had been solved in time
for the meeting ("+") or was still pending at that time ("-"). Fields in this
table are linked to abstracts of publications on solved structures in the
Entrez PubMed system.
Protein name and source [Description]:
This column lists short descriptions of the target proteins, along with the source
organisms.
This column lists the respective prediction team. Team names are linked to
a list of the team members, including crosslinks to abstracts of publications
from team members which are of relevance to their fold-recognition method.
Prediction
This column gives the prediction-ids, which are linked to the original
predictions as submitted
by the authors.
Subset
Some predictors have parsed the target sequence into several subsets or domains,
and one submission might contain predictions for several subsets, which have
been evaluated separately.
This column lists the id-number of the subset, as defined in the original prediction,
for which the respective row of evaluation quantities was calculated.
PDB C D
This column lists the PDB-accession code of the
structural template used for the alignment model, along with the respective
chain-id and the domain-id if applicable. Domains labeled "0" refer to the whole chain,
the domain definitions can otherwise be retrieved from the original predictions.
Fields in this column are linked to MMDB-entries for the respective proteins.
Scomp
This column lists the structure-comparison method, which served as the standard
of truth for the respective row of calculated evaluation quantities. The three
structure-comparison methods used for CASP2 were
VAST,
DALI,
and SSAP.
"NONE" means that the structural template used for the alignment model was not identified as being
similar to the target structure by either of the three methods. Fields in this column are linked
to the original structure-comparison results, if applicable.
For the Alignment model table these are:
Target Id number [Id]:
The fields in this column are linked to the respective entries in the
Protein Data Bank (PDB), where additional information
concerning the processing status or the complete three-dimensional coordinates can be
obtained.
Team
Aron Marchler-Bauer, for CASP2 Organizers