Privateer

Automated detection, building and validation of sugar models starting from X-ray diffraction data

Synthesis: Privateer is a fast software package for the automated detection, model building and validation of carbohydrates. It works directly with structure factors or electron density maps (detection and model building), and uses the same approach to detection already tried and tested by the Buccaneer and Nautilus packages. Privateer incorporates an additional validation tool, privateer-validate, for unbiased asessment of cyclic carbohydrate models. Privateer is part of CCP4, and is being distributed since version 6.5.

[home] [theory] [privateer-validate] [privateer] [source code] [authors] [acknowledgements]

Theoretical background

Conformational validation

Cyclic carbohydrates usually have clear conformational preferences in terms of energy. Enzymes can force sugars into higher-energy conformations in order to achieve catalysis, and these conformations should indeed be kept in mind when modelling a sugar in the -1 site of an enzyme. However, most of the time (i.e. in the rest of the cases) sugars stay in their original lowest-energy conformation, which for pyranoses is either a 4C1 or a 1C4 chair. When a sugar monomer is imported from a dictionary within a graphical model building program such as Coot, its ideal coordinates reflect the lowest-energy conformation, but during the course of refinement they can get distorted and driven away to other conformations such as envelopes or half-chairs. These distortions are usually triggered by:
Weak density gradients
Sometimes there is just partial density for a sugar that is known to be there (e.g. in a High-Mannose glycosylation tree), and it is the crystallographer's decision whether to model it or not. However, atempting to refine a model with such data without the appropriate restraints - torsion restraints can help in this case - will nearly always result in distortion.
Wrong linkage distance specification
Forcing refinement software to restrain a link to a wrong distance will make it fit the model with an offset equal to the deviation from the correct value. This can result just in an increased total puckering amplitude (Q) value if the deviation is not too high, or a complete conformational change for higher values.
Mistaken anomer or handedness
Trying to fit an alpha anomer into density that supports the beta anomer instead will result on a flat ring or a conformational change.

We use the well-stablished Cremer-Pople algorithm (Cremer and Pople, 1975, JACS 97:1354-58) at the core of our conformational analysis. This analysis gives one angle (Phi) and a total puckering amplitude (Q) for furanoses and two angles (Phi, Theta) and Q for pyranoses. The relationship between the different angles and conformations is reflected in the subsequent graphical representation of the conformational parameters:

Conformational landscape for furanose sugars. Atoms without markup are roughly coplanar. This angle assignment implies a clockwise ordering of the ring atoms starting from the oxygen. An opposite-sense naming scheme would imply a 180° shift in Phi.

Cremer-Pople sphere for pyranoses. Chair conformations are located at the poles of the sphere, with the northern hemisphere being the one predominantly occupied by D-pyranoses (due to the configuration of the carbon linked to the highest-ranked in-ring carbon). Half-chairs and Envelopes, which have been omitted in order to produce a clearer figure, are located at Theta=45° and Theta=135° in the northern and southern hemispheres respectively.


The Cremer-Pople algorithm involves the calculation of a uniquely-defined mean plane that satisfies, among other conditions, that the sum of all the vertical distances from the ring atoms to that plane is zero. The stereochemistry can also be easily worked out by calculating equivalent displacements for those atoms relevant to it. After that, Privateer-validate compares the reported conformation and stereochemistry to those calculated for the idealised, lowest-energy structure of the studied sugar.

For a deeper nomenclature check, you can also use the pdb-care server at glycosciences.de (maintained by Dr Thomas Lütteke, Justus-Liebig University Giessen).

Approach to detection

To be completed after the publication of Privateer-detect.

Program description

Privateer-validate

 This validation software will analyse monosaccharide moieties from a PDB file, using a database of three-letter codes, stereochemistry and conformations derived from the wwPDB database. It will also output linear descriptions of any glycans found (N- or O-glycans).
The program takes as input a set of structure factors in MTZ (amplitudes or intensities) or CIF (amplitudes only) formats and a structure in PDB format, and determines the conformation and stereochemistry for all the monosaccharides it's able to find in the structure file. Additionally, a Real Space Correlation coefficient is computed against an omit map calculated excluding the sugars from phase calculations. The program will also produce a visual checklist with the conflicting sugar models in the form of Scheme and Python scripts for use with Coot.

To use them: coot --script privateer-results.py or coot --script privateer-results.scm

Blue map: 2mFo-DFc map. Pink map: mFo-DFc omit map.

Running Coot with these scripts also automatically turns on torsion restraints for real space refinement. This means that there is a better chance of keeping the sugars in their original low-energy conformation if they have not deviated too much from it.

Privateer-validate offers an easy-to-use graphical interface that runs under CCP4i2, and which produces richer reports than those available from its text-based version. The input data window looks like this:

Here is a look at the report window for a dataset where furanoses and pyranoses coexist:

And finally, here is an interactive piece on the glycosylation tree structure reports. You may place your mouse over a sugar to reveal the three-letter code and conformation:

Man←α6−Man←β4−GlcNac← β4−GlcNac ← β−ASN61

GlcNac ← β−ASN211

Man←β4−GlcNac←β4−GlcNac ← β−ASN252

GlcNac←β4−GlcNac ← β−ASN315

Man←α2−Man←α6−(Man←α2−Man←α3−)Man←α6−(Man←α2−Man←α3−)Man←β4−GlcNac←β4−GlcNac ← β−ASN322


If you choose to rebuild your structure with Coot after running Privateer, a visual guided tour of the reported issues will be automatically loaded. Torsion restraints will also be automatically turned on.

Command line usage

prompt% privateer-validate -pdbin file.pdb -mtzin file.mtz -colin-fo F_i24,SIGF_i24

Parameter list

-pdbin [file.pdb]
Compulsory argument: input PDB file to examine. If no CRYST1 card is present, Privateer will still process it but please bear in mind that it may be missing important contacts described by crystallographic symmetry.

-mtzin [file.mtz] or -cifin [file.cif]
An MTZ file containing amplitudes or intensities or a CIF file containing amplitudes. If intensities are supplied, Privateer will run the CCP4 program 'ctruncate' in order to produce amplitudes. Please ensure that CCP4 software is within reach (i.e. included in the PATH variable). Privateer-validate is compatible out of the box with most of the CIF structure factor files downloaded from the PDB. Compatible with miniMTZ files from CCP4i2.
-colin-fo [F,SIGF]
Columns containing F and SIGF, e.g. FOBS,SIGFOBS or F,SIGF without any spaces in between characters. If not supplied, Privateer will try to guess the path using the most commonly used choices.

-codein [SUG]
Three-letter code (those defined by the wwPDB) for the target sugar. If this parameter is not provided, Privateer will scan and score all monosaccharides present in it's database.

-showgeom
Bond lengths, angles and torsions are reported for each ring.

-mtzout [file.mtz]
Optional: output map coefficients to an MTZ file. A name for the file must be provided.

-radiusin [value in angstroems]
Optional: provide a radius for the calculation of the mask around the target sugar. Default value is 1.5 Angstroems.

-mode [normal|ccp4i2]
Optional: run mode. Default is normal. Under ccp4i2 mode, Privateer will generate miniMTZ files, less text output and XML data containing HTML descriptions of any glycosylation trees found, and all the validation data.

Privateer

 To be published very soon.

Source code

 Privateer is Free Software. You can access and/or checkout it's source code at CCP4forge (Bazaar repository, stable versions) or Google Code repository (Subversion repository, experimental stuff that may not work).

Authors

 Jon Agirre and Kevin Cowtan
York Structural Biology Laboratory
Department of Chemistry, The University of York
Wentworth Way, YO10 5DD, England.
Last modified: December 1st, 2014.

Bugs can be reported to jon.agirre (at) york.ac.uk

Acknowledgements

 A number of people have shaped the future of Privateer with their ideas. In no particular order: Eleanor Dodson, Jacob Sřrensen, Keith Wilson, Gideon Davies, Christian 'Bones' Roth, Wendy Offen, Thomas Lütteke, Carme Rovira, Jose A. Cuesta Seijo, Stuart McNicholas, Antoni Planas, Marcelo Guerin, Dan Wright, Glyn Hemsworth, Andrew Thompson, Marcin Wojdyr, Saioa Urresti, Paul Emsley and Robbie Joosten.

[Back to top]