ConSurf 2005: the projection of evolutionary
conservation scores of residues on protein structures
Meytal Landau, Itay Mayrose1, Yossi Rosenberg, Fabian Glaser2, Eric Martz3,
Tal Pupko1and Nir Ben-Tal*
Department of Biochemistry, George S. Wise Faculty of Life Sciences, Tel Aviv University, Ramat Aviv 69978, Israel,
1Department of Cell Research and Immunology, George S. Wise Faculty of Life Sciences, Tel Aviv University,
Ramat Aviv 69978, Israel,2European Bioinformatics Institute, Wellcome Trust Genome Campus, Cambridge,
CB10 1SD, UK and3Department of Microbiology, University of Massachusetts, Amherst, MA 01003, USA
Received February 5, 2005; Accepted March 3, 2005
Key amino acid positions that are important for main-
taining the 3D structure of a protein and/or its func-
tion(s), e.g. catalytic activity, binding to ligand, DNA
constraints. Thus, the biological importance of a
residue often correlates with its level of evolution-
ary conservation within the protein family. ConSurf
(http://consurf.tau.ac.il/) is a web-based tool that
automatically calculates evolutionary conservation
scores and maps them on protein structures via a
user-friendly interface. Structurally and functionally
important regions in the protein typically appear as
patchesof evolutionarily conserved residues that are
Bayesian method for scoring conservation, which is
more accurate than the maximum-likelihood method
steps in the calculation can now be controlled by a
number of advanced options, thus further improving
the accuracy of the calculation. Moreover, ConSurf
version 3.0 also includes a measure of confidence
for the inferred amino acid conservation scores.
The degree to which an amino acid position is recessive to
substitutions is strongly dependent on its structural and func-
tional importance. An amino acid that plays an essential role,
ary conservation is often indicative of the importance of the
position in maintaining the protein’s structure and/or function.
ConSurf is a web server for mapping the level of evolution-
based on the evolutionary relations among the protein and
its homologs and the probability of residue replacement as
reflected in amino acid substitution matrices (2,3). The scores
are subsequently translated into a discrete coloring scale that
is used to project them on a known 3D structure of one of the
homologous proteins. The server is implemented in a user-
friendly interface that enables scientists from the experimental
biology as well as the bioinformatics communities to explore
to identify structurally and functionally important positions.
We provide here a brief review of ConSurf with emphasis
on the new features that were added recently.
A short description of the methodology is provided here and a
under ‘OVERVIEW’, ‘QUICK HELP’ and ‘FAQ’.
A flow chart, describing the ConSurf protocol, is presented
in Figure 1. The minimal input requirement for ConSurf is a
four-letter PDB (4) code and the relevant chain identifier of
the query protein. Alternatively, a user-provided protein struc-
ture in the form of a PDB file can be uploaded. Using the 3D
structure of the query protein as an input, the following steps
are automatically carried out by ConSurf:
(i) The amino acid sequence is extracted from the PDB file.
(ii) Homologous sequences in the SWISS-PROT database (5)
are searched and collected using PSI-BLAST (6).
(iii) A multiple sequence alignment (MSA) of these sequences
is computed using CLUSTAL W (7).
*To whom correspondence should be addressed. Tel: +972 3 640 6709; Fax: +972 3 640 6834; Email: email@example.com
ª The Author 2005. Published by Oxford University Press. All rights reserved.
The online version of this article has been published under an open access model. Users are entitled to use, reproduce, disseminate, or display the open access
version of this article for non-commercial purposes provided that: the original authorship is properly and fully attributed; the Journal and Oxford University Press
only in part or as a derivative work this must be clearly indicated. For commercial re-use, please contact firstname.lastname@example.org
Nucleic Acids Research, 2005, Vol. 33, Web Server issueW299–W302
(iv) A phylogenetic tree is reconstructed based on the MSA,
using the neighbor-joining algorithm (8) as implemented
in the Rate4Site program (3).
(v) Position-specific conservation scores are computed using
the empirical Bayesian (2) or maximum-likelihood (3)
(vi) The continuous conservation scores are divided into a dis-
crete scale of 9 grades for visualization purpose. Grade 1
contains the most variable positions and is colored tur-
quoise; grade 5 contains intermediately conserved posi-
tions and is colored white; and grade 9 contains the most
conserved positions and is colored maroon.
(vii) The nine-color conservation grades are projected onto the
3D structure of the query protein.
The sensitivity and selectivity of the search for homologous
proteins [step (ii) above] can be controlled by adjusting the
number of PSI-BLAST iterations, the PSI-BLAST E-value
cut-off and the maximum number of sequences extracted
from PSI-BLAST (6). As an alternative to this automatic
search, the server accepts a user-provided MSA. In such a
case, steps (ii) and (iii) in the outline protocol are skipped.
After the calculation begins, ConSurf produces a status page
indicating the computation parameters along with the different
stages of the server activity. The main result of a ConSurf
calculation is under the link ‘View ConSurf Results with Pro-
tein Explorer’, which leads to the graphic visualization of the
query protein, color coded by conservation scores, through the
Protein Explorer interface (9). The continuous conservation
scores of each of the amino acid positions are available under
the link ‘Amino Acid Conservation Scores’, along with the
color grades and additional data. The script command for
viewing the 3D structure of the query protein, color coded
by conservation scores, is available under the link ‘RasMol
coloring script source’. This file can be downloaded and used
locally with the RasMol program (10), thus producing the
same color-coded scheme generated by the server. A PDB
file, in which the conservation scores are specified in the
temperature (B) factor field, can be downloaded through the
link: ‘The PDB file updated with the conservation scores in
the tempFactor field’. Thus, any 3D protein viewer, such as the
RasMol program (10), which is capable of presenting the B
factors, is suitable for mapping the conservation scores on the
The ConSurf output also includes links to the PSI-BLAST
results, the homologous sequences along with a link to their
SWISS-PROT entry page, the MSA and the phylogenetic tree
used in the calculation.
As an example, we provide in Figure 2 the main output of a
ConSurf run of the Kcsa potassium-channel (11), a transmem-
brane protein from Streptomyces Lividans. Kcsa is a homotet-
ramer with a 4-fold symmetry axis about its pore. The ConSurf
calculations demonstrate the high level of conservation of the
pore region as compared with the rest of the protein. The pore
architecture provides the unique stereochemistry which is
required for efficient and selective conduction of potassium
ions (11). The biological importance of this stereochemistry is
reflected by a strong evolutionary pressure to resist amino acid
replacements in the pore. In contrast, the regions that surround
the pore and face the extracellular matrix are highly variable.
Figure 1. A flow chart of a ConSurf calculation.
Figure 2. A ConSurf analysis of the Kcsa potassium channel. The tetrameric
using a space-filled model. The amino acids are colored by their conservation
grades using the color-coding bar, with turquoise-through-maroon indicating
variable-through-conserved. Amino acid positions, for which the inferred
conservation level was assigned with low confidence, are marked with light
yellow. The potassium ion at the channel pore is colored green. Conservation
on the homotetrameric structure. The run was carried out using PDB code
1bl8 (11) and default ConSurf parameters. The picture was generated using
MOLSCRIPT (21) and Raster3D (26).
W300 Nucleic Acids Research, 2005, Vol. 33, Web Server issue
NEW ADDITIONS AND IMPROVEMENTS
An empirical Bayesian method to score conservation
The heart of the ConSurf server is the calculation of the con-
servation scores of each amino acid position. In the previous
version of the server, the maximum-likelihood method (3) was
used as the default option to this end. Recently, we showed
that an empirical Bayesian method can significantly improve
the accuracy of the estimated conservation scores (2). The
empirical Bayesian method is particularly superior to the
maximum-likelihood method when the number of homo-
logous sequences analyzed is small (2). The new method is
now integrated in ConSurf as the default option. The usage of
the maximum-likelihood method is still available under the
‘Method’ pull-down menu.
Estimation of the reliability of the inferred
An amino acid position that is conserved across all homo-
logous sequences will always be assigned with the highest
conservation grade. Yet, there is a difference if the conserva-
tion score is inferred based on a small MSA of, for example,
4 sequences, or based on a larger set of 30 sequences. Addi-
tionally, since the conservation calculation for positions with
a lot of gaps is based on a fewer number of sequences, the
conservation score for these positions will be less reliable
than positions that have no gaps. The reliability of the con-
servation computation is not only determined by the number
of sequences in each position but also by the evolutionary
distances between the sequences, the phylogenetic tree
topology and the evolutionary process.
One of the most important new features in ConSurf version
3.0 is the inclusion of a measure of the confidence of each
of the inferred position-specific conservation scores. The
measure is calculated using the empirical Bayesian method,
as explained in (12) and at http://consurf.tau.ac.il/ under
‘OVERVIEW’. In short, it is based on a confidence interval
that is defined by the lower and upper quartiles: the 25th and
75th percentiles of the inferred distribution of conservation
scores, respectively. It gives the 50% confidence interval and
also indicates the dispersion of each of the estimated scores.
The confidence interval is usually large in positions with a
small number of sequences, thus indicating a low level of
support in the inferred conservation scores for these positions.
When the number of sequences is large, the confidence inter-
val is small, and the point score estimates are more assured.
Amino acid positions, associated with confidence intervals
that are too large to be trustworthy, are marked in the output
files of the server and highlighted (in pale yellow) on the 3D
structure of the protein (Figure 2).
Models of amino acid substitutions
The inference of the evolutionary conservation scores relies
on a specified probabilistic model of amino acid replace-
ments (3). The JTT matrix (13) was used to this end in
the previous version of ConSurf. In version 3.0, we expanded
the utility of ConSurf to support additional models of sub-
stitution for nuclear DNA-encoded as well as non-nuclear
DNA-encoded proteins. The model of substitution can be
chosen from the ‘Model of substitution for proteins’ pull-
down menu, which is available in the ‘Advance Options’
section of the ConSurf main interface. The JTT (13),
Dayhoff (14) and WAG (15) matrices are suitable for nuclear
DNA-encoded proteins. The WAG matrix has been inferred
from a large database of sequences comprising a broad range
of protein families, and is thus suitable for distantly related
amino acid sequences (15). The mtREV (16) and cpREV (17)
matrices are suitable for mitochondrial and chloroplast DNA-
encoded proteins, respectively. Examples that demonstrate
the influence of using the different matrices on the calcula-
tions are available at http://consurf.tau.ac.il/ under ‘OVER-
VIEW’. The differences between ConSurf calculations using
different matrices tend to be small but not negligible.
User-provided phylogenetic tree
A user-provided phylogenetic-tree (that should be consistent
with the MSA) may be supplied as an additional input. In this
case, steps (ii–iv) in the ‘ConSurf protocol’ (specified above)
are skipped. We note that the accuracy of the conservation
scores calculations relies on the correct reconstruction of the
phylogeny (18). Default ConSurf runs are carried out using
phylogenetic trees that are constructed with the neighbor-
joining algorithm. The new feature enables the users to supply
more accurate trees.
WORK UNDER DEVELOPMENT
We are currently integrating a few more enhancements
to ConSurf. At present, ConSurf uses the neighbor-joining
algorithm as a fast heuristic method to construct phylogenetic
trees. Notwithstanding, the more exhaustive maximum-
likelihood tree-reconstruction method is known to produce
more accurate phylogenetic trees (19), which should increase
the accuracy of the calculated conservation scores (18). We
will integrate the maximum-likelihood-based SEMPHY pro-
gram (20) into ConSurf. This program reconstructs phylogen-
etic trees dramatically faster than other maximum-likelihood
tree-reconstruction methods (20), and can thus be used with
little additional computational cost.
A computational tool will be developed, which will enable
a simultaneous online view of the phylogenetic tree while
analyzing the evolutionary profile of the protein. This inter-
active tool will allow the user to mark specific branches, which
will be used for in-depth ConSurf analyses. For example, the
selection of specific clades (sub-trees) may be used to define
sub-families. The examination of ConSurf analysis of sub-
families may reveal specific characters that are unique to
each of them.
The main output of ConSurf is the projection of the con-
servation scores on the 3D structure of the query protein. In
order to easily generate high-resolution color figures, we will
provide a script command for the MOLSCRIPT program (21)
as an additional output.
A planned enhancement to ConSurf will be the inclusion
of all the visualization results in the header of the PDB file.
The format that will be used to this end will allow an inter-
active offline view of the results using Protein Explorer on the
user machine, exactly as they appear online.
Nucleic Acids Research, 2005, Vol. 33, Web Server issueW301
CONCLUSIONS Download full-text
ConSurf (1) is a web server that automatically calculates
evolutionary conservation scores for each amino acid position
and projects them onto the 3D structure of the protein. Evolu-
tionary trace (ET) (22,23) is a related web server that may
also be used to map conservation scores on the 3D structure.
However, the ET method (23), which was developed for the
identification of class-specific residues, is less accurate than
ConSurf for scoring conservation (3). This may explain why
biologically important regions that were detected using Con-
Surf were overlooked by the ET web server (1). (See http://
consurf.tau.ac.il/, under ‘OVERVIEW’ for details). Other
related web servers, such as MSA3D (9), ProtSkin (24) and
COLORADO3D (25), may also be used to present conserva-
tion scores on protein structures. These web servers use a
consensus approach to infer conservation, which is inferior
to methods, such as the ET and ConSurf’s maximum-
likelihood and empirical Bayesian that explicitly take into
account the phylogeny of the homologous sequences under
study (2,3). Moreover, all the above servers are not fully
automated as ConSurf and require a user-provided MSA.
The new version of ConSurf includes an improved algo-
rithm for scoring evolutionary conservation and provides an
index of confidence in the estimated scores. In addition, while
ConSurf is still easy to use with default options, expert users
can benefit from several advanced options that were added in
order to provide more control over the calculations and so to
increase the accuracy of the results.
The authors are grateful to the Bioinformatics Unit and the
George S. Wise Faculty of Life Sciences at Tel Aviv
University for providing technical assistance and computation
facilities. This study was supported by a Research Career
Development Award from the Israel Cancer Research Fund.
T.P. was supported by a grant in Complexity Science from the
Development of Protein Explorer is supported by a grant to
E.M. from the Division of Undergraduate Education of the US
National Science Foundation. Funding to pay the Open Access
publication charges for this article was provided by Tel Aviv
Conflict of interest statement. None declared.
1. Glaser,F., Pupko,T., Paz,I., Bell,R.E., Bechor-Shental,D., Martz,E. and
Ben-Tal,N. (2003) ConSurf: identification of functional regions in
proteins by surface-mapping of phylogenetic information.
Bioinformatics, 19, 163–164.
2. Mayrose,I., Graur,D., Ben-Tal,N. and Pupko,T. (2004) Comparison of
site-specific rate-inference methods for protein sequences: empirical
Bayesian methods are superior. Mol. Biol. Evol., 21, 1781–1791.
3. Pupko,T., Bell,R.E., Mayrose,I., Glaser,F. and Ben-Tal,N. (2002)
in proteins by surface mapping of evolutionary determinants within
their homologues. Bioinformatics, 18, S71–S77.
4. Berman,H.M., Westbrook,J., Feng,Z., Gilliland,G., Bhat,T.N.,
Weissig,H., Shindyalov,I.N. and Bourne,P.E. (2000) The Protein Data
Bank. Nucleic Acids Res., 28, 235–242.
5. Bairoch,A.andApweiler,R. (1999)TheSWISS-PROT proteinsequence
data bank and its supplement TrEMBL in 1999. Nucleic Acids Res.,
6. Altschul,S., Madden,T., Schaffer,A., Zhang,J., Zhang,Z., Miller,W. and
protein database search programs. Nucleic Acids Res., 25, 3389–3402.
7. Thompson,J.D., Higgins,D.G. and Gibson,T.J. (1994) CLUSTAL W:
improving the sensitivity of progressive multiple sequence alignment
through sequence weighting, position-specific gap penalties and weight
matrix choice. Nucleic Acids Res., 22, 4673–4680.
for reconstructing phylogenetic trees. Mol. Biol. Evol., 4, 406–425.
9. Martz,E. (2002) Protein Explorer: easy yet powerful macromolecular
visualization. Trends Biochem. Sci., 27, 107–109.
10. Sayle,R.A. and Milner-White,E.J. (1995) RASMOL: biomolecular
graphics for all. Trends Biochem. Sci., 20, 374.
11. Doyle,D.A., Morais Cabral,J., Pfuetzner,R.A., Kuo,A., Gulbis,J.M.,
Cohen,S.L., Chait,B.T. and MacKinnon,R. (1998) The structure of the
potassium channel: molecular basis of K+ conduction and selectivity.
Science, 280, 69–77.
12. Susko,E., Inagaki,Y., Field,C., Holder,M.E. and Roger,A.J. (2002)
Testing for differences in rates-across-sites distributions in phylogenetic
subtrees. Mol. Biol. Evol., 19, 1514–1523.
of mutation data matrices from protein sequences. Comput. Appl.
Biosci., 8, 275–282.
14. Dayhoff,M.O., Hunt,L.T., Barker,W.C., Schwartz,R.M. and Orcutt,B.C.
(1978) In Young,C.L. (eds) Atlas of Protein Sequence and Structure.
National Biomedical Research Foundation, Washington, DC.
15. Whelan,S. and Goldman,N. (2001) A general empirical model of
protein evolution derived from multiple protein families using a
maximum-likelihood approach. Mol. Biol. Evol., 18, 691–699.
16. Adachi,J. and Hasegawa,M. (1996) Model of amino acid substitution in
proteins encoded by mitochondrial DNA. J. Mol. Evol., 42, 459–468.
17. Adachi,J., Waddell,P.J., Martin,W. and Hasegawa,M. (2000) Plastid
genome phylogeny and a model of amino acid substitution for proteins
encoded by chloroplast DNA. J. Mol. Evol., 50, 348–358.
18. Mayrose,I., Mitchell,A. and Pupko,T. (2005) Site-specific evolutionary
rate inference: taking phylogenetic uncertainty into account.
J. Mol. Evol., in press.
19. Felsenstein,J. (1996) Inferring phylogenies from protein sequences by
parsimony, distance, and likelihood methods. Methods Enzymol.,
20. Friedman,N., Ninio,M., Pe’er,I. and Pupko,T. (2002) A structural
EM algorithm for phylogenetic inference. J. Comput. Biol., 9,
21. Kraulis,P.J. (1991) MOLSCRIPT: a program to produce both detailed
and schematic plots of protein structures. J. Appl. Cryst., 24,
TGF-beta and related growth factors: implications for site-directed
mutagenesis. Protein Eng., 13, 839–847.
method defines binding surfaces common to protein families.
J. Mol. Biol., 257, 342–358.
24. Deprez,C., Lloubes,R., Gavioli,M., Marion,D., Guerlesquin,F. and
Blanchard,L. (2005) Solution structure of the E.coli TolA C-terminal
domain reveals conformational changes upon binding to the phage g3p
N-terminal domain. J. Mol. Biol., 346, 1047–1057.
25. Sasin,J.M. and Bujnicki,J.M. (2004) COLORADO3D, a web server for
the visual analysis of protein structures. Nucleic Acids Res., 32,
26. Merritt,E.A. and Bacon,D.J. (1997) Raster3D photorealistic molecular
graphics. Methods Enzymol., 277, 505–524.
W302Nucleic Acids Research, 2005, Vol. 33, Web Server issue