Critical Assessment of PRediction of Interactions
First community wide experiment on the comparative
evaluation of protein-protein docking for structure
Hosted By EMBL/EBI-MSD Group
Target 01 Evaluation
Raul Mendez, Leonardo De Maria and Shoshana J. Wodak have
now completed the evaluation of CAPRI Target_01.
The report is available as a HTML version
NOTE ON PARTICIPANT IDENTIFICATION CODES
In all the evaluation notes the entries are labelled with
a unique identifier, e.g. in files such as T01_P34.8.A.fitting.summary.
The identifier P34 is the
identifier given to the Shoshana's group and is NOT
the same as the identifier given to each participant.
For example the identifiers given in the
Final Summary Do not match the Identifiers given to the
NOTES FROM Shoshana
The results are organised in a summary sheet, as well as
more detailed data, organised in directories.
In brief they computed for each predicted interface: 1. The contacts ratio
2. The fraction of correctly predicted interface residues
for the Hpr and Kinase subunits, respectively,
3. The rigid-body rotation angle that needs to be applied
to the predicted Hpr subunit, in order to superimpose it onto
the Hpr in the target, once the predicted and target kinase
subunits have been superimposed,
4. The distance between the geometric centers of the subunits,
5. The number of close contacts in each predicted interface.
6. We also provide the PDB files, with the predicted complex
superimposed onto the target complex. The superposition was
performed only on the kinase subunits, so that the difference
between the Hpr positions , between the predicted and target
complexes should be clearly visible, using rasmol, for example,
provided adequate colouring is used.
The best predictions are those that have a high contacts
ratio, a small rigid body rotation angle, a small distance
between the geometric centres, and a low number of close
Only a few predictions satisfy these criteria, but the
highest contact ratios are as low as 12-17/52-59 or so.
This is nevertheless very encouraging, given that the bound
and unbound kinase molecules, have different conformations,
making correct predictions of the interaction interface
nearly impossible, using procedures that do not incorporate
conformational changes. Not only does the subunit change
conformation (backbone rms 1.96Ć), but the monomers also
move relative to one another (when dimers are superimposed
the backbone rms is 2.42Å). These changes seem to
result mainly from the movement of the C-terminal helix
( from residue 287 on).
When this helix is not used in the superposition of the
kinase monmers, the backbone rms drops to 1.248Å.
We do not provide at this stage, an appreciation or ranking
of the predictions. This will come later.
Also we do have more data (for example rms values) which
we did not include.
Email Problems or Queries to Kim
Henrick, John Tate