BIOINFORMATICS APPLICATIONS NOTE

Vol. 19 no. 18 2003, pages 25002501

DOI: 10.1093/bioinformatics/btg362

ModLoop: automated modeling of loops in

protein structures
Andrs Fiser1, and Andrej Sali2,
1 Department

of Biochemistry and Seaver Foundation Center for Bioinformatics,

Albert Einstein College of Medicine, 1300 Morris Park Ave., Bronx, NY 10461, USA
and 2 Departments of Biopharmaceutical Sciences and Pharmaceutical Chemistry,
California Institute for Quantitative Biomedical Research, Mission Bay Genentech
Hall, 600 16th Street, Suite N472D, University of California, San Francisco,
CA 94143-2240, USA
Received on May 1, 2003; revised on June 23, 2003; accepted on July 8, 2003

ABSTRACT
Summary: ModLoop is a web server for automated modeling
of loops in protein structures. The input is the atomic coordinates of the protein structure in the Protein Data Bank format,
and the specification of the starting and ending residues of one
or more segments to be modeled, containing no more than
20 residues in total. The output is the coordinates of the nonhydrogen atoms in the modeled segments. A user provides
the input to the server via a simple web interface, and receives
the output by e-mail. The server relies on the loop modeling
routine in MODELLER that predicts the loop conformations by
satisfaction of spatial restraints, without relying on a database
of known protein structures. For a rapid response, ModLoop
runs on a cluster of Linux PC computers.
Availability: The server is freely accessible to academic users
at http://salilab.org/modloop
Contact: andras@fiserlab.org

The function of a protein is generally determined by

shape, dynamics and physiochemical properties of its solvent
exposed molecular surface. Likewise, functional differences
among the members of the same protein family are usually a
consequence of the structural differences on the protein surface. In a given fold family, structural variability is a result
of substitutions, insertions and deletions of residues among
members of the family. Such changes frequently correspond
to exposed loop regions that connect elements of secondary
structure in the protein fold. Thus, loops often contribute to
binding sites and determine the functional specificity of a
given protein framework. Consequently, the accuracy of loop
modeling (Oliva et al., 1997; Fine et al., 1986; Xiang et al.,
2002; van Vlijmen and Karplus, 1997; Rapp and Friesner,
1999; Martin and Thornton, 1996) is a major factor determining the usefulness of comparative protein structure models
To

whom correspondence should be addressed.

2500

(Fiser et al., 2002; Al Lazikani et al., 2001) in studying interactions between the protein and its ligands. Loop modeling
can also be useful in refining low- and medium-resolution
structures determined by X-ray crystallography and NMR
spectroscopy.
Prediction of loop conformations by optimization of an
objective function was implemented in MODELLER (Fiser
et al., 2000). The method optimizes the positions of all
non-hydrogen atoms of a loop in a fixed environment. The
optimization relies on a protocol consisting of the conjugate gradient minimization and molecular dynamics simulation
with simulated annealing. The pseudo-energy function contains terms from a molecular mechanics force field as well as
restraints based on statistical distributions derived from known
protein structures. Bonds, angles, some dihedral angles and
improper dihedral angles are restrained by the corresponding terms in the CHARMM-22 potential function (MacKerell
et al., 1998). The mainchain and sidechain dihedral angles
as well as non-bonded atom pairs are restrained by statistical
potentials extracted from many known protein structures (Sali
and Blundell, 1993; Sali and Overington, 1994; Fiser et al.,
2000; Melo and Feytmans, 1997).
Evaluation of the method relied on 40 randomly selected
loops of known structure at each length from 1 to 14 residues.
The accuracy was determined by building loops both in the
native and distorted loop environments because only approximate loop environments are available in real comparative
modeling applications. The errors in loop predictions increase
with loop length and environment distortion (Fig. 1).
The modeling protocol generates a number of independently optimized conformations, starting with random initial
conformations. The final loop prediction is the optimized conformation that has the lowest pseudo-energy score. Currently,
the number of independent optimizations is limited to 300.
Because each individual optimization typically takes a few
minutes of CPU time on a Pentium processor, the ModLoop
web server runs on a cluster of Pentium nodes.

Bioinformatics 19(18) Oxford University Press 2003; all rights reserved.

Automated modeling of loops in protein structures

Fig. 1. The accuracy of loop modeling as a function of environment

distortion (Fiser et al., 2000). Average prediction accuracy is shown
for 40 loops of 4 (open circles), 8 (filled circles) and 12 residues
(filled squares). The root-mean-square (RMS) error of the loops is
calculated for the four mainchain atom types upon the optimal superposition of the whole model on the native structure. The RMS error
upon superposition of only the loop atoms tends to be 1.5 times
smaller for 8-residue loops than the errors shown (Fiser et al., 2000).
The distortion of the environment is measured by the RMS deviation for the three residues flanking both sides of the loop. Error bars
indicate the standard error of the mean.

The ModLoop server requires as input a coordinate file in

the Protein Data Bank (PDB) format, as well as the starting
and ending positions of the loops. The user can specify several
loops as input. These loops will be optimized simultaneously,
which is particularly useful if conformations of multiple interacting loops need to be predicted. Given the rapid decrease in
the prediction accuracy as the loop length increases, the total
number of residues in all selected loops is currently limited
to 20. The resulting coordinate file in the PDB format is sent
back to the user by e-mail.
The ModLoop server is a useful addition to the already available protein structure modeling servers (http://salilab.org/
bioinformatics_resources.shtml). It has been utilized in several applications, including modeling of the active site loop
conformations in dehydrogenases with various substrate specificities (Wu et al., 1999; Vernal et al., 2002), predicting
the conformation of the linker loop in an artificial construct of a circularly permuted cyanovirin protein (Barrientos
et al., 2001), and rationalizing the observed functional impact
of various mutants of the zebrafish winged helix protein
No Soul/Foxi1 (Lee et al., 2003).

REFERENCES
Al Lazikani,B., Jung,J., Xiang,Z. and Honig,B. (2001) Protein
structure prediction. Curr. Opin. Chem. Biol., 5, 5156.

Barrientos,L.G., Campos-Olivas,R., Louis,J.M., Fiser,A., Sali,A.

and Gronenborn,A.M. (2001) 1H, 13C, 15N resonance assignments and fold verification of a circular permuted variant of the
potent HIV-inactivating protein cyanovirin-N. J. Biomol. NMR,
19, 289290.
Fine,R.M., Wang,H., Shenkin,P.S., Yarmush,D.L. and Levinthal,C.
(1986) Predicting antibody hypervariable loop conformations.
II: minimization and molecular dynamics studies of MCPC603
from many randomly generated loop conformations. Proteins, 1,
342362.
Fiser,A., Do,R.K. and Sali,A. (2000) Modeling of loops in protein
structures. Protein Sci., 9, 17531773.
Fiser,A., Feig,M., Brooks,C.L.,III. and Sali,A. (2002) Evolution and
physics in comparative protein structure modeling. Acc. Chem.
Res., 35, 413421.
Lee,S.A., Shen,E.L., Fiser,A., Sali,A. and Guo,S. (2003)
The zebrafish forkhead transcription factor Foxi1 specifies
epibranchial placode-derived sensory neurons. Development, 130,
26692679.
MacKerell,A.D.,Jr, Bashford,D., Bellott,M., Dunbrack,R.L.,Jr,
Evanseck,J.D., Field,M.J., Fischer,S., Gao,J., Guo,H., Ha,S.
et al. (1998) All-atom empirical potential for molecular modleing and dynamics studies of proteins. J. Phys. Chem. B, 102,
35863616.
Martin,A.C. and Thornton,J.M. (1996) Structural families in loops
of homologous proteins: automatic classification, modelling and
application to antibodies. J. Mol. Biol., 263, 800815.
Melo,F. and Feytmans,E. (1997) Novel knowledge-based mean force
potential at atomic level. J. Mol. Biol., 267, 207222.
Oliva,B., Bates,P.A., Querol,E., Aviles,F.X. and Sternberg,M.J.
(1997) An automated classification of the structure of protein
loops. J. Mol. Biol., 266, 814830.
Rapp,C.S. and Friesner,R.A. (1999) Prediction of loop geometries
using a generalized born model of solvation effects. Proteins, 35,
173183.
Sali,A. and Blundell,T.L. (1993) Comparative protein modelling by satisfaction of spatial restraints. J. Mol. Biol., 234,
779815.
Sali,A. and Overington,J.P. (1994) Derivation of rules for comparative protein modeling from a database of protein structure
alignments. Protein Sci., 3, 15821596.
van Vlijmen,H.W. and Karplus,M. (1997) PDB-based protein loop
prediction: parameters for selection and methods for optimization.
J. Mol. Biol., 267, 9751001.
Vernal,J., Fiser,A., Sali,A., Muller,M., Jose,C.J. and Nowicki,C.
(2002) Probing the specificity of a trypanosomal aromatic alphahydroxy acid dehydrogenase by site-directed mutagenesis. Biochem. Biophys. Res. Commun., 293, 633639.
Wu,G., Fiser,A., ter Kuile,B., Sali,A. and Muller,M. (1999) Convergent evolution of Trichomonas vaginalis lactate dehydrogenase
from malate dehydrogenase. Proc. Natl Acad. Sci. USA, 96,
62856290.
Xiang,Z., Soto,C.S. and Honig,B. (2002) Evaluating conformational
free energies: the colony energy and its application to the problem
of loop prediction. Proc. Natl Acad. Sci. USA, 99, 74327437.

2501