Bioinformatics Advance Access originally published online on May 24, 2005
Bioinformatics 2005 21(15):3327-3328; doi:10.1093/bioinformatics/bti511
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Resmap: automated representation of macromolecular interfaces as two-dimensional networks
Department of Biochemistry and Molecular Biology, The University of Kansas Medical Center Kansas City, KS 66160, USA
*To whom correspondence should be addressed.
 |
Abstract |
|---|
 TOP Abstract 1 INTRODUCTION2 PROGRAM DESCRIPTIONREFERENCES |
|---|
Summary: To aid detailed comparison of a large number of macromolecular structures, Resmap imports Protein Data Bank files and represents subunit/domain interfaces as two-dimensional networks.
Availability: http://www.kumc.edu/biochemistry/resmap/
Contact: lswint-kruse{at}kumc.edu
Supplementary information: Default definitions and directions for graphically managing networks are available at the same website.
|
1 INTRODUCTION |
|---|
|
TOPAbstract1 INTRODUCTION2 PROGRAM DESCRIPTIONREFERENCES |
|---|
Many proteins participate in large complexes that cross-communicate through subunit interfaces to execute an overall function. Protein–DNA interfaces may also be important communication avenues (Falcon and Matthews, 2000). These interfaces are often dynamic, changing between functionally relevant protein states. As large numbers of related structures become available, investigators face the challenge of comparing multiple structures/interfaces with reasonable detail. However, for more than 2–3 structures with a few highlighted side chains, structure-based alignments and direct comparison are visually overwhelming. A second comparator—values of C
r.m.s.d.—loses side chain information; and, functionally important changes at a few backbone positions can be lost in the average of a large protein (Swint-Kruse, 2004). To overcome these difficulties, interaction data have been re-cast as two-dimensional networks, in which interface residues serve as nodes and cross-subunit interactions form the edges. This approach has been useful for finding relevant differences and similarities in the protein–protein and protein–DNA interfaces of homologues, monitoring structural changes during dynamics simulations and comparing structures of different functional states for a single protein (Swint-Kruse et al., 2001; Swint-Kruse et al., 2002; Swint-Kruse, 2004). However, these interfaces are laborious to create manually. We have written a computer program to automatically generate network maps and allow them to be interactively changed in real-time.
|
2 PROGRAM DESCRIPTION |
|---|
|
TOPAbstract1 INTRODUCTION2 PROGRAM DESCRIPTIONREFERENCES |
|---|
Resmap is written in C/C++ and runs on Windows 2000 and XP operating systems. Command-line versions are available for Linux and Mac OS-10. Resmap imports Protein Data Bank (PDB) files, calculates atom–atom distances for a chosen pair of subunits and generates networks which can be interactively altered by changing cut-off distances. Final networks are output to postscript files which can be read into Ghostscript (recommended via GSview), and converted to either encapsulated postscript files or windows metafiles, which requires ‘pstoedit’. The relevant websites are: www.cs.wisc.edu/~ghost/, http://www.pstoedit.net/ and www.cs.wisc.edu/~ghost/gsview/. These files can be imported into Microsoft PowerPoint, allowing modifications such as changing line width to show an interaction unique to one structure. In this format, multiple networks can be placed side-by-side for facile comparison of the interfaces. Postscript files record all variable input values (e.g. cut-off distances, definition files) as remarks that can be viewed when opened as text files.
2.1 Atom-type and interaction definitions
Atom-type and interaction definitions are contained in Resmap configuration files (rcf), which may be modified and renamed by the end-user. Resmap automatically opens a default rcf (supplied with the program), but alternate rcf files may be explicitly opened prior to choosing a PDB file.
Default atom definitions (hydrophobic, acceptor, donor, etc.) are largely based on the definitions developed for the program ‘Contacts of Structural Units’ (CSU) (Sobolev et al., 1999). Three changes have been made in the current default data: (1) The CG atom of asparagine has been re-assigned as a ‘neutral’ (type 6), in agreement with the CSU definition for the analogous glutamine atom. (2) The SG atom of cysteine has been re-assigned to atom-type 7 (neutral–donor) (http://www.biochem.ucl.ac.uk/bsm/atlas/); for a disulfide bond, we suggest that this atom-type revert to the CSU assignment of 6. (3) The CA of glycine has been changed to atom-type 7, in agreement with the alpha carbons of other residues. Further, we added the possibility of explicit ion pair interactions by defining positively charged (type 9) and negatively charged (type 10) atoms. A detailed list of the other assignments may be accessed with the Supplementary Data.
Default atom–atom interaction definitions (Supplementary Material) are also based upon those in CSU (Sobolev et al., 1999) but are expanded to include new categories. Specifically, interaction definitions are: HB, hydrogen bond; PH, hydrophobic interaction; AR, aromatic interaction; IP, ion pair; DC, destabilizing contact; and OT, ‘other’, which can generally include van der Waals interactions. Default distance lengths for each interaction can be set in the configuration file. These distances can also be altered within the program to see the interface network change as distances are adjusted. Bond angles are not considered in the currentanalysis.
2.2 Structural input and distance extraction
Structural PDB files are opened within Resmap. Inside the program, any two subunits of interest and a residue range may be chosen. Self versus self networks are allowed and will accommodate domain–domain interfaces, but residue ranges must differ for the network to have meaningful data. Distances are calculated from the PDB X, Y and Z coordinates using the standard algebraic distance equation for all combinations of atom pairs between all combinations of residues pairs from the two chosen subunits.
2.3 Network creation
Networks are plotted on the right-hand side of the Resmap window and may be interactively changed by altering the subunit choice, residue range, distance cut-off lengths for particular interactions or network presentation size. If either the rcf or PDB file is changed, the previous project must be closed and the new version(s) opened. A good general strategy is to first survey plots for the entire interface before narrowing in on specific interactions, so that unexpected long-distance perturbations can be detected.
Networks created by Resmap for LacI and PurR proteins are in very good agreement with those created manually (data not shown) (Swint-Kruse et al., 2001; 2002; Swint-Kruse, 2004). Examples of new networks created by this program and altered in PowerPoint are presented in Figure 1. Hemoglobin was chosen because it provides a classic example of a functionally important, structural change at an interface. In this exercise, we generated networks for all six interfaces in the tetramer for eleven different hemoglobin structures (data not shown) and imported them into PowerPoint in 3 h. Comparison of the networks clearly showed that 1ß1 interfaces do not vary, while other interfaces have significant change (Fig. 1). Networks should direct investigators back to specific regions of the structures for further examination.
|
|
Acknowledgments |
|---|
We thank Cole Stephens (KUMC) for testing the instructions in the Supplementary Material and Dr Todd Holyoak (KUMC) for testing several versions of Resmap. In memoriam: Hiram S. Brown 1918–2005, who encouraged three generations, including both authors, to study science and engineering.
Conflict of Interest: none declared.
Received on January 19, 2005; revised on April 18, 2005; accepted on May 18, 2005
|
REFERENCES |
|---|
|
TOPAbstract1 INTRODUCTION2 PROGRAM DESCRIPTIONREFERENCES |
|---|
Falcon, C.M. and Matthews, K.S. (2000) Operator DNA sequence variation enhances high affinity binding by hinge helix mutants of lactose repressor protein. Biochemistry, 39, 11074–11083[CrossRef][Medline].
Kavanaugh, J.S., et al. (1993) Accommodation of insertions in helices: the mutation in hemoglobin Catonsville (Pro 37 alpha-Glu-Thr 38 alpha) generates a 3(10)
bulge. Biochemistry, 32, 2509–2513[CrossRef][Medline].
Shaanan, B. (1983) Structure of human oxyhaemoglobin at 2.1 Å resolution. J. Mol. Biol., 171, 31–59[ISI][Medline].
Sobolev, V., et al. (1999) Automated analysis of interatomic contacts in proteins. Bioinformatics, 15, 327–332
Swint-Kruse, L. (2004) Using networks to identify fine structural differences between functionally distinct protein states. Biochemistry, 43, 10886–10895[CrossRef][Medline].
Swint-Kruse, L., et al. (2001) Plasticity of quaternary structure: twenty-two ways to form a LacI dimer. Protein Sci., 10, 262–276
Swint-Kruse, L., et al. (2002) Fine-tuning function: correlation of hinge domain interactions with functional distinctions between LacI and PurR. Protein Sci., 11, 778–794
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||