[10:18 21/9/2010 Bioinformatics-btq480.tex] Page: 2624 2624–2625
BIOINFORMATICS APPLICATIONS NOTE
Vol. 26 no. 20 2010, pages 2624–2625
doi:10.1093/bioinformatics/btq480
Structural bioinformatics Advance Access publication August 24, 2010
2Struc: the secondary structure server
D. P. Klose1, B. A. Wallace2 and Robert W. Janes1,∗
1School of Biological and Chemical Sciences, Queen Mary, University of London, Mile End Road, London E1 4NS
and 2Department of Crystallography, Institute of Structural and Molecular Biology, Birkbeck College, University of
London, London WC1E 7HX, UK
Associate Editor: Anna Tramontano
ABSTRACT
Summary: The defined secondary structure of proteins method is
often considered the gold standard for assignment of secondary
structure from three-dimensional coordinates. However, there are
alternative methods. ‘2Struc: The Secondary Structure Server’ has
been created as a single point of access for eight different secondary
structure assignment methods. It has been designed to enable
comparisons between methods for analyzing the secondary structure
content for a single protein. It also includes a second functionality,
‘Compare-the-Protein’ to enable comparisons of the secondary
structure features from any one method to be made within a
collection of nuclear magnetic resonance models, or between the
crystal structures of two different proteins.
Availability: http://2struc.cryst.bbk.ac.uk
Contact: r.w.janes@qmul.ac.uk
Supplementary information: Supplementary data are available at
Bioinformatics online.
Received on June 29, 2010; revised on August 13, 2010; accepted
on August 14, 2010
1 INTRODUCTION
In 1983, Kabsch and Sander (1983) developed what has become
the de facto approach when defining secondary structure, the
defined secondary structure of proteins (DSSP). DSSP uses a single
hydrogen bond energy term to assign eight states of secondary
structure elements (SSEs) to three-dimensional coordinates and is
arguably the golden standard for this task. Although extensively
used, DSSP is not the only tool available and other approaches have
been developed which incorporate different means of defining SSEs.
It has been suggested that DSSP can miss the elucidation of edge
strands (Martin et al., 2005) and has difficulties dealing with low
resolution structures (Martin et al., 2005) and with structures where
only the Cα trace atoms are reported (Labesse et al., 1997). Further
methods have been developed including DSSPcont (Anderson et al.,
2002), STRIDE (Frishman and Argos, 1995), PALSSE (Majumdar
et al., 2005), P-SEA (Labesse et al., 1997), KAKSI (Martin et al.,
2005), STICK (Taylor, 2001) and XTLSSTR (King and Johnson,
1999), the latter created for deriving secondary structure from threedimensional
coordinates to match comparable data determined from
circular dichroism spectra. These other methods meet specific needs
that are not optimally addressed by DSSP; all have merit, and as
such broaden the means by which SSEs can be identified.
The aim of the 2Struc server is to provide easy access to the
information produced by a variety of these SSE assignment tools.
∗To whom correspondence should be addressed.
Currently, eight methods are available on the server, covering all
atom approaches such as DSSP, DSSPcont (an extension of DSSP
using more detailed hydrogen bonding energies) and STRIDE,
vector/line segment methods such as STICK and PALSSE and
distance/angle methods such as P-SEA, KAKSI and XTLSSTR.
Citation references to each of the methods are provided on the
server and we request all users to cite these original articles. The
output of each method is presented numerically in the form of total
secondary structure content as well as for individual residues in
a protein’s sequence and on a graphical display of the protein’s
three-dimensional structure using a Jmol (http://www.jmol.org/)
representation. In addition to method comparisons, differences
within nuclear magnetic resonance (NMR) models and between
two x-ray structures are possible using the ‘Compare-the-Protein’
feature.
2 2Struc FEATURES
The 2Struc server comprises two functionalities: 2Struc and
Compare-the-Protein. A screenshot of the front page is available
as Supplementary Material.
2.1 2Struc analyses
2Struc analyses are compatible with protein atomic coordinates
either by specifying the PDB (Berman et al., 2007) code or by
uploading a coordinates file in PDB format containing ATOM
specifiers with or without SEQRES data. Once uploaded, the file
is presented to the ‘control panel’ the point from which the user
can select which structures are to be analyzed, the tools to be used
and the analyses to run. The five types of analyses available are as
follows:
Protein Structure Summary provides a list of composite chains
along with their lengths. 2Struc calculates the percentage SSE either
according to the original method’s classifications or a reduced threestate
representation: helix (encompassing H, I, G), sheet (E) and
‘other’ (-) (all remaining elements), where H, I, G and E are from
their DSSP definitions (Kabsch and Sander, 1983). The percentage
of each structure element is presented in a tabular form allowing
easy comparisons to be drawn. The three-state SSE assignments
for each method are compared to DSSP using several commonly
levied metrics including the percentage similarity, the more stringent
Matthews correlation coefficient (MCC) (Matthews, 1975) and the
MCC squared. This summary information can be downloaded as a
plain text file.
Structure Summary by Chain is comparable to the previous
analysis, but is now on a per chain basis. For each of the
chains analyzed, the original method’s classifications and threestate
percentages are displayed (as above); however, rather than
© The Author(s) 2010. Published by Oxford University Press.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/
by-nc/2.5), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
[10:18 21/9/2010 Bioinformatics-btq480.tex] Page: 2625 2624–2625
2Struc
having a statistical comparison, the user can view the different
assignments on a three-dimensional representation of the structure
using an embedded Jmol viewer.
Multiple Structure Alignment presents all the reduced threestate
assignments for each individual residue aligned in parallel to
the protein sequence. For clarity, residues that are in helices are
background-colored red and those in beta strands green. A majority
vote consensus is also included that is similarly colored.
Original Structure Alignment presents SSE alignments for each
method that produces more than a three-state output. The majority
consensus is not included here as the algorithms share few
commonly defined elements; however, the structures are colored
in a similar way to those of the previous analysis approach.
Sequence Structure Alignment presents three state and, where
available, original SSE output assignments aligned to the sequence
for each method, and similarly colored as before.
2.2 Compare-the-Protein analyses
Compare-the-Protein is the second functionality. It allows the user
to examine, for each method, structural variation either within a
collection of NMR models or between two x-ray crystallography
structures. Comparisons can be either in reduced three-state
format or in original output assignments where determined. For
comparisons within NMR models, the SSE assignments for each
model are aligned to the sequence and colored in the same manner
as was used in 2Struc. X-ray structure comparisons are between
chains. The user can either use the default of the entire chain or
specify the start and end position of each chain. The chains can be
aligned by SSE using one of two different methods: (i) the tm-align
algorithm (Zang and Skolnick, 2005) or (ii) minimizing the number
of differences between the assigned SSEs retaining the chains as
contiguous components. The SSE assignments are aligned to each
of the protein sequences and a ‘D’ is indicated in the comparison
output line where a difference exists between them.
3 APPLICATIONS
2Struc offers a means of comparing specific SSEs for a given threedimensional
protein structure as defined by eight different methods.
Compare-the-Protein enables comparisons of SSE assignments to
be made between two or more crystal/NMR/model structures for
any chosen method. Envisioned applications for the 2Struc and
Compare-the-Protein tools are as follows:
As a means of determining the ‘consensus’ SSEs within a protein
as defined from a variety of methods.
For comparison between multiple chains, either for identical
subunits within a protein quaternary structure or identical components
within the asymmetric unit of a crystal structure, to determine
possible areas of differences and the cause of such differences.
As a tool for developing new methods for determining secondary
structure based on different structural biology techniques such as
circular dichroism, Raman and infrared spectroscopies.
For visual display of differences between homologous proteins or
wild-type and mutant protein structures.
As a means of identifying proteins that have unusual SSE features
that might be of interest for further structural examination.
To identify differences in SSE assignments between apo- and
ligand-bound structures of a protein near to an active site.
For visualization of differences in SSEs that arise from binding
of an allosteric effector molecule at a site distal to the active site.
For comparing the SSEs of protein structures prepared under
different crystallization conditions.
For comparing the SSE assignments of a solution (NMR) and
crystal structure of the same protein.
For comparing the SSEs of a model structure with an experimentally
determined cognate or homologous structure.
To enable comparisons of structurally homologous proteins that
are not sequence homologous to determine how closely their SSE
assignments align.
4 CONCLUSION
The 2Struc server provides two functionalities; 2Struc, which
generates SSE assignments for protein structures for up to eight
different methods enabling easy analyses of similarities and
differences between them, and Compare-the-Protein, which allows
comparisons and highlighting of differences between the SSEs
generated within a series of NMR models or between two x-ray
structures. 2Struc is unique in providing summaries of percentage
secondary structure content for reduced three-state data and original
SSE output assignments where generated. The server is freely
available to all users at http://2struc.cryst.bbk.ac.uk.
ACKNOWLEDGEMENTS
We thank Dr Lee Whitmore for testing and improvements, Dr
Andrew Miles and Ben Woollett for comments on the manuscript
and Dr David Houldershaw for computing technical support, all
from Birkbeck College. We also thank Prof. William R. Taylor from
the National Institute of Medical Research for supplying the source
code for STICK.
Funding: U.K. Biotechnology and Biological Sciences Research
Council (BBSRC) [BB/F010362 to R.W.J. and BB/F010346 to
B.A.W.].
Conflict of Interest: none declared.
REFERENCES
Anderson,C.A.F. et al. (2002) Continuum secondary structure captures protein flexibility.
Structure, 10, 175–184.
Berman,H. et al. (2007) The worldwide Protein Data Bank (wwPDB): ensuring a single,
uniform archive of PDB data. Nucleic Acids Res., 35, D301–D313.
Frishman,D. and Argos,P. (1995) Knowledge-based protein secondary structure
assignment. Proteins Struct. Funct. Genet., 23, 566–579.
Kabsch,W. and Sander,C. (1983) Dictionary of protein secondary structure: pattern
recognition of hydrogen-bonded and geometrical features. Biopolymers, 22,
2577–2637.
King,S.M. and Johnson,W.C. (1999) Assigning secondary structure from protein
coordinate data. Proteins Struct. Funct. Genet., 35, 313–320.
Labesse,G. et al. (1997) P-SEA: a new efficient assignment of secondary structure from
C alpha trace of proteins. Comput. Appl. Biosci., 13, 291–295.
Majumdar,I. et al. (2005) PALSSE: a program to delineate linear secondary structural
elements from protein structures. BMC Bioinformatics, 6, 202.
Martin,J. et al. (2005) Protein secondary structure assignment revisited: a detailed
analysis of different assignment methods. BMC Struct. Biol., 5, 17.
Matthews,B.W. (1975) Comparison of the predicted and observed secondary structure
of T4 phage lysozyme. Biochim. Biophys. Acta, 405, 442–451.
Taylor,W.R. (2001) Defining linear segments in protein structure. J. Mol. Biol., 310,
1135–1150.
Zhang,Y. and Skolnick,J. (2005) TM-align: a protein structure alignment algorithm
based on TM-score. Nucleic Acids Res., 33, 2302–2309.
2625