|"What's New In CCP4 6.0"||CCP4 Newsletter 44 Summer 2006|
CCP4, CSE Department, CCLRC Daresbury Laboratory, Warrington WA4 4AD
CCP4 version 6.0 is the most recent major release of the CCP4 software suite and was made in February 2006. A subsequent patch release 6.0.1 was made in June to fix a number bugs, however there were no major changes to functionality.
The full list of changes to the suite since the previous major version (5.0) can be found in the CHANGES file distributed with the core suite. As the CHANGES file focuses mainly on technical changes, the aim of this article is to give a quick tour of the new and updated features of the latest release in a more user-friendly manner.
The article has the following outline:
The current release sees the rollout of a new integrated download and install mechanism for the software, which is intended to vastly simplify the otherwise complication task of selecting packages for download via a simple interactive web interface. It also provides mechanisms for easy installation of the packages once they have been transferred to the user's system, using installers appropriate to the operating system in question - see the article on the new system elsewhere in this newsletter.
Using these mechanisms means that it is possible to have CCP4 and the other packages described below working on a machine in just a few minutes. Alternatively, old-fashioned style FTP download is also still provided. All the downloads can be accessed via the CCP4 Downloads Pages.
For release 6.0 it was decided to break the distribution of the suite up into the following "packages":
The breakdown is for practical purposes to do with availability, dependencies and rate of change of the various pieces of software - inclusion or exclusion from the "core" was a pragmatic decision and not an indication of the "quality" or "importance" of a particular package. In this context the "core CCP4 package" essentially refers to the the minimal CCP4 suite containing the CCP4 libraries plus the standard programs, the user interface CCP4i, the Clipper software, and some general "third-party" libraries.
Each of the new and updated packages are described briefly in the sections below.
PDBCUR is a utility program for manipulating coordinate files, similar to PDBSET but already with a large number of useful complementary functions (for example selecting one model from a file which contains many models).
Additional functions new for CCP4 6.0 include:
The options marked with * are also available via the CCP4i interface, in the Edit PDB File task in the Coordinate Utilities module (note that you must select the pdbcur option in order to access the new functions).
MATTHEWS_COEF is traditionally used to estimate the number of molecules in the unit cell based on solvent content. In CCP4 6.0 it has been updated to additionally output the Kantardjieff and Rupp resolution-based probabilities for each possible number of molecules (Protein Science 12 1865-1871 (2003)), providing an additional tool for estimating the contents of the cell.
The corresponding CCP4i task Cell Content Analysis in the Molecular Replacement module has also been updated accordingly.
The RCSB PDB_EXTRACT program has been upgraded since CCP4 5.0, to version 1.7 in release 6.0 and more recently version 2.0 in 6.0.1. PDB_EXTRACT can automatically extract data from program log files and other output as the structure determination progresses, and thus help to ease the burden of the final deposition process (see the article in the previous newsletter on this and other RCSB PDB Tools). PDB_EXTRACT is also accessible from within CCP4i via the Data Harvesting Management Tool in the Validation and Deposition module.
For more information about PDB_EXTRACT and other RCSB PDB Tools, see http://sw-tools.pdb.org/.
The core CCP4 has a number of other updates to existing programs, most significantly:
There are a number of new programs in the core CCP4 package in 6.0:
Each of these is described in more detail in the following sections.
SUPERPOSE is a secondary-structure based structural alignment program. It is an alternative to the least-squares fitting method used in LSQKAB - SUPERPOSE uses the secondary structure features to perform the initial fit, followed by alignment of the protein backbone Cα atoms.
The program takes two coordinate files as input and outputs the transformation matrix required to map one (the "moving" coordinate set) onto the other (the "fixed" set). It also gives a per-residue listing of the quality of the match, and identifies the secondary structure.
SUPERPOSE can be run from within CCP4i by selecting the ...Secondary Structure Matching option in the Superpose molecules task (in the Coordinate Utilities module). It is also used as the structure alignment engine within both CCP4mg (under Applications -> Superpose proteins) and Coot (under Calculate -> SSM Superpose...).
BP3 is a heavy atom refinement and phasing program which uses a multivariate likelihood approach that has been reported to perform well in tests compared with other substructure refinement programs (see the references in the CRANK/BP3 article in an earlier newsletter). It can be applied to various types of experiment including single- and multiple isomorphous replacement (SIR and MIR), single- and multiple-wavelength anomalous dispersion (SAD and MAD) and various combinations thereof (i.e. SIRAS and MIRAS experiments).
Another advantage BP3 offers over CCP4's workhorse substructure refinement and phasing program MLPHARE, is that where MLPHARE forces the user to treat MAD phasing experiments with a "pseudo-MIR" approach (effectively selecting one wavelength as the "native" and then treating the other wavelengths as "derivatives"), BP3 treats each anomalous dataset equally and so has a better treatment of the phase errors.
BP3 can be run from CCP4i via the dedicated Bp3 - Phasing task in the Experimental Phasing module, or as part of the automated CRANK task in the same module (CRANK is described further on in this article).
More information on BP3 can also be found at the BP3 webpage.
CHAINSAW is a molecular replacement preparation utility that mutates a template PDB file using a previously-generated sequence alignment provided by the user. CHAINSAW preserves parts of the template which are conserved in the sequence alignment (and which are therefore more likely to be conserved in the target structure), and prunes back other parts which are not conserved (and which are therefore less likely to reflect the target).
CHAINSAW offers a number of different pruning methods (back to gamma atom, back to beta atom, or back to last "common" atom) and preserves more atoms than for example in a polyalanine model. It can accept sequence alignments in a variety of different formats (including ClustalW, PIR and Blast among others), which has the advantage that the user can optimise the alignment using their favourite methods before feeding it into CHAINSAW.
CHAINSAW can be run from CCP4i from the Create Search Model task in the Molecular Replacement module.
PIRATE is a statistical phase improvement program - crudely, it is a replacement for the DM program. PIRATE uses phases from a previously-solved known structure (called the reference structure) as input to the phase improvement process, and doesn't require any a priori knowledgement of the solvent content of the target. PIRATE is run from the command line using the cpirate executable, or else through the Run Pirate task in the Density Improvement module of CCP4i.
In tests conducted by the program's author PIRATE performed better than DM in all but one case. It should be noted however that the program's performance depends critically on the quality of the Hendrickson-Lattmann (H-L) coefficients output from the phasing program used to generate the initial phases (as these give an indication of the error in the input phases). The current version has been "tuned" to work best with H-L coefficients output from PHASER but is known to also work well with the output from SOLVE. Using the output from MLPHARE will generally work less well, and tests with BP3 have not been carried out yet.
PIRATE can also make automatic use of non-crystallographic symmetry (NCS), if it is given a file of heavy atom coordinates.
For typical structures without any unusual features the choice of reference structure is not critical provided that it is relatively large - therefore the CCP4 distribution includes the structure factors for 1AJR in .na4 format (so-called "ascii-MTZ" - to convert to MTZ proper use the NA4TOMTZ program, or run the Convert to MTZ... task in the Reflection Data Utilities module). For structures with more unusual features (for example metalloproteins or RNA-complexes) it would be worth trying different reference structures which more accurately reflect the features of the target - in this case PIRATE includes a option to evaluate multiple reference structures (check Use evaluation mode in the PIRATE interface, or specify the -evaluate option if running cpirate).
The program MAKEREFERENCE is used to generate the reference input for PIRATE (use the cmakereference executable, or the Make Pirate Reference task in the Density Improvement module of CCP4i). MAKEREFERENCE requires both coordinates and structure factors for the reference structure, and outputs new coordinates and structure factors for input into PIRATE. If the computer has an internet connection then this data can be automatically downloaded from the PDB archive at the EBI.
The PIRATE program is built on top of a set of crystallographic software libraries called Clipper. In addition to PIRATE and MAKEREFERENCE, CCP4 6.0 and later also contain a number of other Clipper-based utilities that perform small tasks such as generating maps and comparing phases. The utilities can be accessed from CCP4i via tasks in the Clipper Utilities module, and the CCP4i documentation gives a list of the available programs.
Information about PIRATE, Clipper and the related software can be found in Kevin Cowtan's webpages.
PDB_MERGE is a useful jiffy for combining models from different sources, for example: constructing a complex from its components, constructing a new model of a protein using domains from several other models, or adding in separately generated symmetry mates. PDB_MERGE has just two modes (MERGE/NOMERGE), which control whether chains are merged at the same time as merging the data in the files.
CCP4i is the CCP4 graphical user interface system. The current version (released in CCP4 6.0.1) is 22.214.171.124, and the major changes have already been reported in a previous newsletter article. These developments are summarised in the following sections.
The new core tools are illustrated in figure 1 and include:
Figure 1: a map of the new core CCP4i tools and features
There are two major new interfaces:
CRANK is a suite of programs for automated macromolecular structure solution. Currently Crank supports SAD, SIR and SIRAS experiments and makes use of various new and existing programs, including BP3, SHELX and various CCP4 programs.
CRANK starts from scaled and merged data and allows the automatic solution of macromolecular structures up to the point of density modification. However it also has a "transculent box" design, intended to help teach novice users about the various programs used in crystallography.
The CRANK interface is part of CCP4i's Experimental Phasing module. A more detailed article about CRANK appeared in a previous newsletter; alternatively, visit the CRANK webpage for further information.
The SHELX_CDE task interfaces to the SHELX programs, specifically SHELX C, D and E. The task takes either SCALEPACK format reflection files (note that the SCALA task can now output SCALEPACK-style files that are suitable for input into SHELX and SOLVE) or MTZ files containing intensities (preferred) or structure factor amplitudes as input.
The task can be used to run the programs in a "pipeline" fashion from data preparation through heavy atom site location to density modification and hand determination, and generates useful plots from the output of each program.
Like CRANK, the SHELX_CDE interface is also part of CCP4i's Experimental Phasing module. Note that the SHELX programs themselves must be obtained directly from the SHELX website.
PHASER is a maximum likelihood-based phasing program. The current version distributed by CCP4 (PHASER 1.3) has methods for molecular replacement, however functions for experimental phasing are also under development. One of PHASER's strengths is that its scoring function gives a more accurate estimate of the quality of its molecular replacement solutions; another is that it can search for solutions using ensembles made up of many possible search models, with each model's contribution to the ensemble weighted by estimates of its similarity to the target. It also allows searching for multiple molecules in multiple spacegroups.
PHASER's functionality can be accessed in a variety of ways - it allows each step in the MR process to be run independently, alternatively there is an "automatic" mode which will run through the whole process without user intervention. PHASER can also be run via a CCP4i task interface in the Molecular Replacement module.
PHASER depends upon the CCTBX crystallographic software libraries, and these must also be installed before building PHASER from source (they are not required for binary installations). For more information about CCTBX see the CCTBX pages on SourceForge. For more information about PHASER, see the PHASER webpages.
CCP4mg is the official CCP4 molecular graphics package, and has focused so far on structure analysis and high-quality presentation graphics (including movies). CCP4mg is highly compatible with the CCP4 environment, and has an interface with a similar look-and-feel to CCP4i.
CCP4mg is built around the idea of data objects (the raw data that is loaded into the program, such as coordinates or reflections) and display objects (the representations of that data in the graphics window). A single data object can have many associated display objects, which can represent all or just some of the data (for example just the active site, a single chain or the whole molecule) in a variety of different ways (for example as ball-and-stick, ribbons, spheres or surfaces) and can reflect different properties of the data (for example secondary structure elements, solvent accessibility or electrostatic potential). Multiple display objects are easily generated and combined to quickly build complex representations.
Some examples taken from the CCP4mg website are shown in figures 2 thtough 4 below. These and other images showing the capabilities of the program can be seen in the CCP4mg Gallery, and in the examples in the 10-minute tutorials.
|Figure 2: Displaying a surface coloured by electrostatic potential||Figure 3: Active site in 1DFR, showing the local environment||Figure 4: Displaying an electron density map|
|* Images taken from the CCP4mg website.|
CCP4mg has support for generating molecular surfaces (including transparent surfaces), and has various display modes for situations like protein-RNA complexes and ligand binding to DNA. The program can also display maps (for example to show electron density), either directly from a CCP4 map file or generated "on the fly" from an MTZ file containing structure factor amplitudes and phases. In addition it has tools for model superposition (via SUPERPOSE), generating symmetry and packing diagrams, and for examining the local environment for part of a protein model.
There is also functionality for adding captions and legends, for customising colours, and for rendering images at high-resolutions suitable for publication.
For more information see the CCP4mg website.
Coot is a platform for semi-automated model building and validation tools, and as such its functionality complements that in CCP4mg. The validation tools in Coot (including "dynamic" Ramachrandran plots as shown in figure 5, and geometry and B-factor graphs) allow the user to quickly focus in on areas of the model that are a poorly fitted, and once problems are identified there are a number of tools to help fix them (for example, rotamer fitting as shown in figure 6, residue mutation, loop fitting and real-space refinement amongst others).
Coot also offers tools to search for waters and for other "blobs" of unmodelled density, and to then model them for example by placing atoms in the density, or by allowing the user to pull ligands from the REFMAC monomer library and "drop" them into the model. Coot also has the option to run rounds of REFMAC refinement from within the program, as well as a large number of other tools for model completion not described here.
For further information about Coot visit the Coot webpages.
|Figure 5: Example of Coot's dynamic Ramachandran plots||Figure 6: Example of rotamer fitting in Coot|
|* Images taken from the Coot website.|
Having accurate values for the anomalous scattering factors f' and f'' is essential for the success of anomalous phasing methods. CHOOCH is able to determine the values of these parameters from the raw fluorescence spectra. Currently there is no CCP4i interface to run CHOOCH, however this will be addressed in a future release.
An article on CHOOCH (in Word format) appeared in a previous newsletter, or alternatively additional information is available at the CHOOCH webpages.
Version 6.0 of the CCP4 software suite is available for download and use free of charge for academic and non-profit use, upon the completion of a valid licence agreement. The academic licence is included with the suite, or else can be found via our licence pages, however please note that the licence for version 5.0 is not valid for version 6.0. Organisations wishing to use the software for commercial purposes should contact CCP4 to obtain a commercial licence - this information is also available via our licence pages.
Whenever possible we have endevoured to provide both source code and precompiled binaries for all the programs and packages on all the platforms that we have access to. This includes various flavours of Linux and some older UNIXes, as well as Microsoft Windows and Mac OS-X. Our intention is to give users of the software as many options as possible for installation.
The current major exceptions are CCP4mg and Coot, both of which are only available from the CCP4 website as binary packages. This is for practical reasons only: building these programs from source code is not a trivial task and depends on the system setup and on the level of experience of the installer. For those who do wish to take up the challenge, the source code can be obtained from the appropriate websites: http://www.ysbl.york.ac.uk/~ccp4mg/download/ for CCP4mg and http://www.ysbl.york.ac.uk/~emsley/coot/ for Coot.
As part of the life cycle of any software release, bugs are continually being discovered and fixed. Fixes to problems are usually posted on the Problems Pages, and typically take the form of a "workaround" (suggested steps to avoid the bug) or as fixes to the program source code. In this case the fixes are normally distributed as "patches" (a patch is a file containing instructions on how to modify another file, which can be read by the UNIX patch program). It is always a good idea to check the Problems Pages before reporting problems.
CCP4 6.0 and later includes a new utility called patch_ccp4.sh, which on UNIX systems will automatically check for available patches to the release, download and apply them. If your installation was from source code then you still need to recompile any updated programs, however this will also apply fixes to things like CCP4i which do not need to be (re)compiled.
Depending of the lifespan of the release and the number of fixes accumulated since the last release, patch releases such as 6.0.1 are occasionally produced. These updates should not add any new functionality. It is likely that at some stage there will be a 6.0.2 patch release of CCP4 6.0.
Since the release of CCP4 6.0, CCP4 staff and associated developers have been visiting groups at various labs in the UK (and in one case further afield) in order to demonstrate the new features of the software and talk to scientists about using CCP4. Feedback from these visits has been very positive, so if you feel that your group might benefit from a visit by CCP4 staff and developers then please contact us at email@example.com.
CCP4 staff are also attending a number of international conferences, with exhibition stands at the ECM and ACA conferences amongst others, and will be happy to demonstrate the software and to answer questions about it. Please see the CCP4 courses pages for the most up to date information.
Beyond CCP4 6.0 there are a number of other current and forthcoming programs and projects associated with CCP4. Many of these are linked from either the CCP4 Projects Page or the Prerelease Page.
Some of the programs that are now currently available outside of the CCP4 6.0 include:
MrBump: Molecular replacement with Bulk Model Preparation
MrBump is an automated molecular replacement pipeline, with emphasis on generating a variety of search models. It uses various "helper" applications (e.g. CHAINSAW), bioinformatics tools (e.g. FASTA) and on-line databases (e.g. the PDB) to generate models, and then uses MOLREP, PHASER and REFMAC for the MR steps.
In favourable cases MrBump has given a "one-button" MR solution; in less favourable cases it can still be used to generate search models for further investigation.
MrBump has a CCP4i task interface but can also be run standalone. It can be obtained via http://www.ccp4.ac.uk/MrBUMP/ and will be included in the next major release of CCP4.
PISA: Protein Interfaces Surfaces and Assemblies
PISA is an interactive tool for the exploration of macromolecular (proteins and DNA/RNA) interfaces, prediction of probable quaternary structures (assemblies) and database searches of structurally similar interfaces and assemblies.
PISA is currently available as a webservice hosted by the MSD-EBI: http://www.ebi.ac.uk/msd-srv/prot_int/pistart.html. It is also planned that a future release of CCP4 will feature a standalone version of PISA.
iMOSFLM is the new interface to the MOSFLM data processing program. It is intended to replace the current dated X11-based interface with a more intuitive and user-friendly graphical interface that will also work on Microsoft Windows.
The GUI is now available for public download from http://alf1.mrc-lmb.cam.ac.uk/~geoff/mosflm/.
POINTLESS does a number of things: it can be used to determine the Laue group and spacegroup from a set of unmerged reflections; it can score possible indexing schemes for a merged or unmerged dataset against a merged dataset; and it can be used to reindex or change the spacegroup of a reflection file.
POINTLESS can be obtained from the CCP4 Prerelease Page.
BUCCANEER is a new Clipper application that performs statistical chain tracing by identifying connected alpha-carbon positions using a likelihood-based density target. It is essentially a model building program, and follows on very naturally from the output of experimental phasing steps and in particular from PIRATE. Tests by the program's author suggest that it is complementary to ARP/wARP, since BUCCANEER is relatively insensitive to resolution but very sensitive to the quality of the phases and the estimates of their errors.
BUCCANEER can be obtained via http://www.ysbl.york.ac.uk/~cowtan/.
The CCP4 project is a collaborative effort and continues to thrive through generous contributions of time, energy and software from members of the UK and internation PX communities. Unfortunately time and space do not permit the acknowledgement here of all these valuable contributions, however a list of acknowledgements is included in the current release, and acknowledgements for the specific developments described in this article are given below:
The images for CCP4mg and Coot were taken without permission (and thus with apologies) from the relevant websites. PJB would also like to apologise to anyone who has been missed from these acknowledgements.
The CCP4 suite is maintained, developed and released by the CCP4 group in the Computational Science and Engineering Department at CCLRC Daresbury Laboratory, and comprises Charles Ballard, Peter Briggs, Maeri Howard, Ronan Keegan, Francois Remacle, Dan Rolfe, Norman Stein and Martyn Winn.
The CCP4 project is supported by the BBSRC, by income from commercial distribution of the software, and by CCLRC Daresbury Laboratory. CCP4 would also like to thank the many people past and present who support the project, both with their time and with their contributions to the software suite itself - without which the project would not be able to exist.
Peter Briggs, July 5th 2006