ZDOCK and RDOCK Performance in CAPRI Rounds
3, 4, and 5
Kevin Wiehe,1Brian Pierce,1Julian Mintseris,1Wei Wei Tong,2Robert Anderson,1Rong Chen,1and
1Bioinformatics Program, Boston University, Boston, Massachusetts
2Department of Biomedical Engineering, Boston University, Boston, Massachusetts
results of our ZDOCK and RDOCK algorithms in
Rounds 3, 4, and 5 of the protein docking challenge
CAPRI. ZDOCK is a Fast Fourier Transform (FFT)-
based, initial-stage rigid-body docking algorithm,
and RDOCK is an energy minimization algorithm
for refining and reranking ZDOCK results. Of the 9
targets for which we submitted predictions, we
attained at least acceptable accuracy for 7, at least
medium accuracy for 6, and high accuracy for 3.
These results are evidence that ZDOCK in combina-
tion with RDOCK is capable of making accurate
predictions on a diverse set of protein complexes.
We present an evaluation of the
© 2005 Wiley-Liss, Inc.
Key words: ZDOCK; RDOCK; CAPRI; protein dock-
ing; scoring function; blind test
CAPRI is an experiment for members of the protein
docking community to test their algorithms for accurately
predicting the 3-dimensional (3D) structures of protein
complexes given only the independently solved structures
of their constituents.1We participated in the first 2 rounds
of CAPRI with the ZDOCK algorithm and achieved some
successful predictions.2,3ZDOCK is an initial-stage dock-
ing algorithm that utilizes a Fast Fourier Transform (FFT)
search algorithm and a 3-term energy function consisting
of shape complementarity, electrostatics, and desolvation
energy.4The shape complementarity term is based on
pairwise atom distance calculations and was shown to
complementarity terms.5In addition, we have developed a
benchmark of protein complexes in order to extensively
test our docking algorithms that is freely available for the
protein docking community.6The benchmark has recently
been updated and now contains over 80 protein complex
Before Round 3 of CAPRI, we developed the RDOCK
algorithm to refine and rerank the top 2000 predictions of
ZDOCK with an energy minimization protocol utilizing
tially improve docking results when used with 3 versions
of ZDOCK that employ various combinations of the 3
energy terms in their respective scoring functions.9Round
3 of CAPRI marked the first time RDOCK was imple-
mented into our docking approach in a blind test setting.
Through the results from the latest 3 rounds of CAPRI,
we have demonstrated that our approach of using ZDOCK
initial-stage docking combined with RDOCK refining and
reranking is highly successful for making accurate protein
MATERIALS AND METHODS
Our approach to Rounds 3, 4, and 5 of CAPRI involved
refining and reranking ZDOCK’s top 2000 initial predic-
tions with RDOCK, followed by clustering these results
and then visually examining the lowest energy members of
each cluster. Biological information from the literature
was incorporated in our approach by either preventing
contacts by using distance filters on predictions after
RDOCK. For the last 2 rounds of CAPRI, we also applied a
slightly different approach to clustering than in the earlier
rounds. Because some test cases in the past were more
accurately predicted with different versions of ZDOCK, we
sought a method to combine several independent runs of
ZDOCK, each with 2000 predictions. Specifically, we ran 2
versions of ZDOCK, one with just the pairwise shape
complementarity term and the other with all 3 energy
each with RDOCK. Finally we clustered these predictions
along with predictions from the ZDOCK run with all 3
energy terms. The entire docking approach applied to the
latest rounds of CAPRI is outlined in detail as follows.
In most targets submitted, biological information was
used on some level to discourage contacts between certain
residues in the ZDOCK predictions. Blocking is accom-
plished by assigning all atoms of the residues to be
prohibited from the putative binding site a special atom
is executed. In comparison to a contact filter on ZDOCK
output, blocking is advantageous because predictions with
undesirable contacts are unlikely to appear in the top 2000
Grant sponsor: National Science Foundation; Grant numbers: DBI-
0078194, DBI-0133834, and DBI-0116574.
ment of Biomedical Engineering, Boston University, 44 Cummington
Street, Boston, MA 02215. E-mail: email@example.com
Received 16 January 2005; Accepted 31 January 2005
PROTEINS: Structure, Function, and Bioinformatics 60:207–213 (2005)
© 2005 WILEY-LISS, INC.
We believe ZDOCK and RDOCK represent a successful
first step in the solution to the docking problem. Our perfor-
algorithms, as part of a comprehensive general docking
strategy, can correctly identify at least 1 accurate structure
out of 10 submissions for a large majority of blind-test cases.
The diversity of protein complexes on which our approach is
successful only furthers the idea that combining initial-stage
docking techniques such as ZDOCK and RDOCK that can
identify potential near-native structures with more sophisti-
cated refinement methods that can account for side-chain
lead to great progress in the protein docking field. We look
their extremely helpful efforts.
1. Janin J, Henrick K, Moult J, Ten Eyck L, Sternberg MJE, Vajda S,
Vasker I, Wodak SJ. CAPRI: A Critical Assessment of PRedicted
Interactions. Proteins 2003;52:2–9.
2. Me ´ndez R, Leplae R, De Maria L, Wodak SJ. Assessment of blind
predictions of protein–protein interactions: current status of dock-
ing methods. Proteins 2003;52:51–67.
3. Chen R, Tong WW, Mintseris J, Li L, Weng ZP. ZDOCK predic-
tions for the CAPRI challenge. Proteins 2003;52:68–73.
5. Chen R, Weng ZP. A novel shape complementarity scoring func-
tion for protein–protein docking. Proteins 2003;51:397–408.
6. Chen R, Mintseris J, Janin J, Weng ZP. A protein–protein docking
benchmark. Proteins 2003;52:88–91.
7. Mintseris J, Wiehe K, Pierce B, Anderson RJ, Chen R, Janin J,
Weng Z. Protein–Protein Docking Benchmark 2.0: an update.
8. Brooks BR, Bruccoleri RE, Olafson BD, States DJ, Swaminathan
S, Karplus M. CHARMM: a program for macromolecular energy,
minimization, and dynamics calculations. J Comput Chem 1983;4:
9. Li L, Chen R, Weng ZP. RDOCK: Refinement of rigid-body protein
docking predictions. Proteins 2003;53:693–707.
10. Chen R, Weng Z. A novel shape complementarity scoring function
for protein–protein docking. Proteins 2003;51:397–408.
11. Chen R, Li L, Weng Z. ZDOCK: an initial-stage protein-docking
algorithm. Proteins 2003;52:80–87.
12. Pierce B, Tong W, Weng Z. M-ZDOCK: a superior grid-based
approach for Cnsymmetric multimer docking. Bioinformatics
13. Poschl E, Mayer U, Stetefeld J, Baumgartner R, Holak TA, Huber
R, Timpl R. Site-directed mutagenesis and structural interpreta-
tion of the nidogen binding site of the laminin gamma1 chain.
EMBO J 1996;15:5154–5159.
14. DeLano WL. The PyMOL molecular graphics system. San Carlos,
CA: DeLano Scientific; 2002.
15. Takagi J, Yang Y, Liu JH, Wang JH, Springer TA. Complex
between nidogen and laminin fragments reveals a paradigmatic
beta-propeller interface. Nature 2003;424:969–974.
16. Bower MJ, Cohen FE, Dunbrack RL Jr. Prediction of protein
side-chain rotamers from a backbone-dependent rotamer library:
a new homology modeling tool. J Mol Biol 1997;267:1268–1282.
17. Mintseris J, Weng Z. Atomic contact vectors in protein–protein
recognition. Proteins 2003;53:629–639.
17a.Graille M, Zhou CZ, Receveur-Brechot V, Collinet B, Declerck N,
van Tilbeurgh H. Activation of the LicT transcriptional antitermi-
nator involves a domain swing/lock mechanism provoking mas-
sive structural changes. J Biol Chem 2005;280:14780–14789.
18. Bressanelli S, Stiasny K, Allison SL, Stura EA, Duquerroy S,
Lescar J, Heinz FX, Rey FA. Structure of a flavivirus envelope
glycoprotein in its low-pH-induced membrane fusion conforma-
tion. EMBO J 2004;23:728–738.
19. Shimon LJ, Bayer EA, Morag E, Lamed R, Yaron S, Shoham Y,
Frolow F. A cohesin domain from Clostridium thermocellum: the
crystal structure provides new insights into cellulosome assembly.
20. Schwede T, Kopp J, Guex N, Peitsch MC. SWISS-MODEL: an
automated protein homology-modeling server. Nucleic Acids Res
21. Mechaly A, Yaron S, Lamed R, Fierobe HP, Belaich A, Belaich JP,
Shoham Y, Bayer EA. Cohesin–dockerin recognition in cellulo-
some assembly: experiment versus hypothesis. Proteins 2000;39:
22. Mechaly A, Fierobe HP, Belaich A, Belaich JP, Lamed R, Shoham
Y, Bayer EA. Cohesin–dockerin interaction in cellulosome assem-
bly: a single hydroxyl group of a dockerin domain distinguishes
between nonrecognition and high affinity recognition. J Biol Chem
23. Miras I, Schaeffer F, Beguin P, Alzari PM. Mapping by site-directed
mutagenesis of the region responsible for cohesin–dockerin interac-
tion on the surface of the seventh cohesin domain of Clostridium
24. Carvalho AL, Dias FM, Prates JA, Nagy T, Gilbert HJ, Davies GJ,
Ferreira LM, Romao MJ, Fontes CM. Cellulosome assembly
revealed by the crystal structure of the cohesin-dockerin complex.
Proc Natl Acad Sci USA 2003;100:13809–13814.
25. Goldberg J, Huang HB, Kwon YG, Greengard P, Nairn AC, Kuriyan
J. Three-dimensional structure of the catalytic subunit of protein
26. Tanaka J, Ito M, Feng J, Ichikawa K, Hamaguchi T, Nakamura M,
Hartshorne DJ, Nakano T. Interaction of myosin phosphatase
target subunit 1 with the catalytic subunit of type 1 protein
phosphatase. Biochemistry 1998;37:16697–16703.
27. Hirano K, Phan BC, Hartshorne DJ. Interactions of the subunits
of smooth muscle myosin phosphatase. J Biol Chem 1997;272:
28. Zhao S, Lee EY. A protein phosphatase-1-binding motif identified
by the panning of a random peptide display library. J Biol Chem
29. Toth A, Kiss E, Herberg FW, Gergely P, Hartshorne DJ, Erdodi F.
Study of the subunit interactions in myosin phosphatase by
surface plasmon resonance. Eur J Biochem 2000;267:1687–1697.
30. Terrak M, Kerff F, Langsetmo K, Tao T, Dominguez R. Structural
basis of protein phosphatase 1 regulation. Nature 2004;429:780–
31. Krengel U, Dijkstra BW. Three-dimensional structure of endo-1,4-
beta-xylanase I from Aspergillus niger: molecular basis for its low
pH optimum. J Mol Biol 1996;263:70–78.
32. Furniss CS, Belshaw NJ, Alcocer MJ, Williamson G, Elliott GO,
Gebruers K, Haigh NP, Fish NM, Kroon PA. A family 11 xylanase
from Penicillium funiculosum is strongly inhibited by three wheat
xylanase inhibitors. Biochim Biophys Acta 2002;1598:24–29.
33. Bray MR, Clarke AJ. Identification of a glutamate residue at the
active site of xylanase A from Schizophyllum commune. Eur
J Biochem 1994;219:821–827.
34. Ko EP, Akatsuka H, Moriyama H, Shinmyo A, Hata Y, Katsube Y,
Urabe I, Okada H. Site-directed mutagenesis at aspartate and
glutamate residues of xylanase from Bacillus pumilus. Biochem J
35. Wakarchuk WW, Campbell RL, Sung WL, Davoodi J, Yaguchi M.
Mutational and crystallographic analyses of the active site residues
36. Miao S, Ziser L, Aebersold R, Withers SG. Identification of
glutamic acid 78 as the active site nucleophile in Bacillus subtilis
xylanase using electrospray tandem mass spectrometry. Biochem-
37. Sansen S, De Ranter CJ, Gebruers K, Brijs K, Courtin CM,
Delcour JA, Rabijns A. Structural basis for inhibition of Aspergil-
lus niger xylanase by Triticum aestivum xylanase inhibitor-I.
J Biol Chem 2004;279:36022–36028.
38. Eghiaian F, Grosclaude J, Lesceu S, Debey P, Doublet B, Treguer E,
Rezaei H, Knossow M. Insight into the PrPC3PrPSc conversion
ZDOCK AND RDOCK PERFORMANCE IN CAPRI ROUNDS