Practical Crank

Structure Refinement for Robots, Automatons, and Graduate Students.

Steven Ness

Copyright (c) 2004 - Steven Ness - Leiden University Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation; with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts. A copy of the license is included in the section entitled "GNU Free Documentation License".

2004

Table of Contents

1. Introduction

1.1: What is Crank?
1.2: What crystallographic programs can Crank use?
1.3: What kind of crystallographic experiments can be performed in Crank?

2. Installation

2.1: How do I install Crank?

3. Running Crank

3.1: How do I run Crank?
3.2: What do all the fields in the Crank interface mean?
3.2.1: Title
3.2.2: Type of Experiment
3.2.3: Procedure
3.2.4: Programs: (Substructure Determination, Refinement and Phasing, Density
3.2.5:MTZ in:
3.2.6:CuKalpha wavelength:
3.2.7:f' and f''
3.2.8:wavelength:
3.2.9:Columns:
3.2.10:MTZ out
3.2.11:Substructure:
3.2.11.1: Atom name
3.2.11.2: Num. Atoms:
3.2.12:Macromolecule - Num. Residues or Num. Nucleotides.
3.2.13:B-Factor
3.2.14:Solvent Content
3.2.14.1: Calculate Bfator and Solv. Content
3.2.15:Number of atoms in asymmetric unit:
3.2.15.1: Calculate from Macromolecule contents

4. Interpreting Results

4.1: Crank output .mtz file
4.2: Crank log file
4.2.1: Crank header and references
4.2.2: Conversion to Scalepack files
4.2.3: Extraction and Scaling
4.2.4: Superimpose substructure on model
4.2.5: Individual Program Logfiles
4.2.5.1: DREAR output
4.2.5.2: CRUNCH2 output
4.2.5.3: SHELXC/D output
4.2.5.4: BP3 output
4.2.5.5: DM output
4.2.5.6: SHELXE output

5. Chapter 5: What to do if you encounter problems

5.1: Problems with the Crank GUI
5.2: Problems with the Crank engine
5.3: Problems with CRUNCH2
5.4: Problems with BP3

6. Chapter 6: Crank Internals

6.1: Crank XML
6.2: Crank directory structure
6.3: Crank Tcl / Crank CCP4i

7. Chapter 7: More Information

7.1: Crank
7.2: CRUNCH2
7.3: BP3
7.4: SHELX
7.5: CCP4/CCP4i

8. Chapter 8: References

Bibliography

Chapter 1. Introduction

Table of Contents

1.1: What is Crank?
1.2: What crystallographic programs can Crank use?
1.3: What kind of crystallographic experiments can be performed in Crank?

1.1: What is Crank?

Crank is a program to automate protein crystal structure determination, when starting from good quality reflection data. It indeed does not do the heavy lifting itself, but instead uses many high quality crystallographic programs, giving them properly formatted input files, and interpreting their output. It is designed to be a transparent box, allowing automation of the structure determination process in easier cases, and allow the user to run and cross compare results obtained from different techniques in difficult cases.

1.2: What crystallographic programs can Crank use?

Crank currently has interfaces for CRUNCH2, BP3, the DREAR suite, SHELX(C/D/E), DM and SOLOMON. In addition, Crank has internal interfaces to many of the CCP4 suite of programs, including the CCP4i user interface, SFTOOLS, TRUNCATE, WILSON, MTZ2VARIOUS, CAD, SFTOOLS, MTZMADMOD, F2MTZ and others. Support for many other programs is currently being added to Crank.

If you are a developer and would like Crank to interface to your program, please contact us. It's a simple and quick process, especially with the help of the developer.

1.3: What kind of crystallographic experiments can be performed in Crank?

Currently, Crank has support for SAD, SIR and SIRAS experiments. Support for MAD and MIR(AS) is being added.

Crank takes merged intensity data from a CCP4 MTZ file. It is important for the direct methods programs used to have intensity data, as a small, but sometimes crucial bit of information is lost in the conversion from Intensities (I's) to Amplitudes (F's). However, soon, Crank will allow for import of both Amplitude and/or Intensity data.

From this data, Crank can first determine either the anomalously diffracting or heavy atom substructure, depending on the experiment type. These tasks can be performed by either CRUNCH2 or SHELXC/D. It can also take user entered coordinates, if these are previously known. It then can refine these coordinates with BP3 or MLPHARE, and produce phase estimates. These phase estimates can then be density modified by either SOLOMON, DM or SHELXE.

Chapter 2. Installation

Table of Contents

2.1: How do I install Crank?

2.1: How do I install Crank?

1) Download the crank-XXXX-0.8.tgz file for your required platform(s) from: http://www.bfsc.leidenuniv.nl/software/crank
2) Untar the crank tar file
```
tar xvzf crank-XXXXX-0.8.0.tgz
							
```

3) Add the crank bin directory to your path

csh or tcsh:


setenv PATH $PATH\:/directory/to/crank/bin

bash:


export PATH="$PATH:/directory/to/crank/bin"

4) Set the environment variable CRANK to the main crank directory

csh or tcsh:


setenv CRANK "/directory/to/crank"

bash:


export CRANK="/directory/to/crank"

5) Install the CCP4i interface from the file "crank.tar.gz" in crank/ccp4i directory with the following steps. (a) Start ccp4i (as the CCP4 system administrator) (b) From the "System Administration" menu, choose the "Install Tasks" option. (c) From the "Task archive" panel, select the "crank.tar.gz" located in the crank/ccp4i directory. (d) Click on "Apply" and restart ccp4i - the crank task should appear at the end in the "Program List" and "Experimental Phasing" module!
6) Install the BP3 CCP4i interface. Complete the same steps as listed in #5, replacing the .tar.gz with "bp3.tar.gz"
7) Test Crank In order to fully test your Crank engine installation, you can type the following commands. The Crank job that runs will take from 15 min. to over an hour to complete, depending on your processor.
```
cd $CRANK/test/dna360
crank dna360.input.xml > & log
tail -f log
								
```
If the job completes without error, your Crank engine install is complete.
8) Run Crank engine If you are already running the CCP4i interface, stop all running instances, and type the command:
```
ccp4i
							
```
Then, using the main CCP4i menu on the far left hand side of the interface, select "Program List", then scroll down to "Crank" and click it.
9) Tutorials We have prepared a SAD tutorial for users. All input files have been uploaded to the Crank website, please visit it for the latest versions at: http://www.bfsc.leidenuniv.nl/software/crank

Chapter 3. Running Crank

3.1: How do I run Crank?

Crank is most easily run through its CCP4i interface. To start the Crank CCP4i interface type the command:


ccp4i

Then, using the main CCP4i menu on the far left hand side of the interface, select "Program List", then scroll down to "Crank" and click it.

3.2: What do all the fields in the Crank interface mean?

The main fields in the Crank interface are described below. Optional fields in subtasks are not yet included.

3.2.1: Title

A short descriptive title for the experiment to appear in the CCP4i task window

3.2.2: Type of Experiment

What type of crystallographic experiment was performed at the xray source. Currently SAD, SIR and SIRAS experiments are supported. Support for MAD and MIR(AS) is is development.

3.2.3: Procedure

What crystallographic programs to run on the input reflection data. You can currently just run Substructure Determination, or you can combine it with both Refinement and Phasing and Density Modification.

3.2.4: Programs: (Substructure Determination, Refinement and Phasing, Density

Modification) Specifies the programs to be used for these respective steps. By choosing different programs, different parts of the crystallographic parameter and phase space can be explored, sometimes leading to results where running just one set of parameters would fail.

3.2.5:MTZ in:

The name of the input MTZ file with merged reflection data. Currently, the data must contain both Amplitude and Intensity measurements to allow for the best performance of all the programs in the pipeline. In the future, either Intensity or Amplitude data may be used.

3.2.6:CuKalpha wavelength:

If the experiment was performed at the Copper Kalpha emission wavelength of 1.54A, Crank can estimate the f' and f'' values from values from the CCP4 file $CCP4/lib/data/atomsf.lib.

3.2.7:f' and f''

If the experiment was performed at a synchrotron, please enter the f' and f'' values as measured from a fluorescence scan.

3.2.8:wavelength:

The wavelength of X-ray radiation used to illuminate the crystal.

3.2.9:Columns:

In the case of SAD: The F+ and F- along with I+ and I- columns to be used. In the case of SIR: For both the native and the derivative, the F and I columns to be used. In the case of SIRAS: For the native, the F and I columns to be used and for the derivative both the F+ F- and I+ I- columns to be used.

3.2.10:MTZ out

The filename of the output .mtz file. A default value will be automatically determined from the input file name.

3.2.11:Substructure:

First, the anomalously scattering and/or heavy atom substructure must be found. In order to do this, only the number and type of heavy atom must be entered. In addition, you can investigate other options in the substructure panel in the CCP4i interface.

3.2.11.1: Atom name

The CCP4 compliant atom name in the substructure. This name must be found in the file $CCP4/lib/data/atomsf.lib in the CCP4 distribution to be considered valid. In the case of two or more anomalously diffracting heavy atoms, enter the atom type for the atom contributing more to the anomalous scattering, and try increasing the atom number slightly. For example, in the case of lysozyme, often both the chlorine and sulphur atoms can be found by using an atom type of sulphur.

3.2.11.2: Num. Atoms:

The approximate number of heavy atoms in the substructure. Although it is important to try to estimate this number accurately, it is better to be a little on the high side, as occupancy refinement in BP3 can often determine the validity of lesser occupied atomic sites.

3.2.12:Macromolecule - Num. Residues or Num. Nucleotides.

In order to calculate several important properties of your macromolecule, please enter the number of residues and/or number of nucleotides. This information is used mainly in two places, first, along with the .mtz file to calculate the approximate B-factor and solvent content below. Second, it is used to calculate the macromolecular contents, which are used to calculate E(A) values in the process of determining the substructure.

3.2.13:B-Factor

The average atomic B-factor.

3.2.14:Solvent Content

The solvent content of your crystal. The program can provide a guess (described below) which should be examined by the user to see if it is reasonable.

3.2.14.1: Calculate Bfator and Solv. Content

If an .mtz file has been read in and the number of residues and nucleotides has been entered, the CCP4 program WILSON can be run by pressing this button to calculate approximate B-factor and solvent content values.

3.2.15:Number of atoms in asymmetric unit:

In order to produce better estimates of E(A) values, the DREAR suite requires the approximate number of light atoms in the asymmetric unit.

3.2.15.1: Calculate from Macromolecule contents

If the number of residues and nucleotides has been entered, the approximate number of atoms in the asymmetric unit can be calculated by pressing this button.

Chapter 4. Interpreting Results

Table of Contents

4.1: Crank output .mtz file
4.2: Crank log file
4.2.1: Crank header and references
4.2.2: Conversion to Scalepack files
4.2.3: Extraction and Scaling
4.2.4: Superimpose substructure on model
4.2.5: Individual Program Logfiles
4.2.5.1: DREAR output
4.2.5.2: CRUNCH2 output
4.2.5.3: SHELXC/D output
4.2.5.4: BP3 output
4.2.5.5: DM output
4.2.5.6: SHELXE output

4.1: Crank output .mtz file

In the Crank CCP4i interface, you specify the Crank output MTZ file, this file is created in the directory you specify. If you have troubles finding it, look in your CCP4 project directories.

This .mtz file contains both all the input columns that Crank knows about, copied from your input .mtz file. In addition, it contains all the final phase estimates, figures of merit and Hendrickson-Lattman parameters, as estimated by the various programs in the crystallographic pipeline.

4.2: Crank log file

The Crank log file contains output not only from Crank, but from all the programs in the pipeline, therefore, this file can grow very large. Although we apologize for the size, we think that it is important for the user to read and understand the information in the logfile.

To this end, we are currently developing tools in the Crank suite to parse the logfile information from all the various crystallographic tools and to convert it into a format amenable both for users to examine and for programs to use to make decisions on the direction of the pipeline, depending on program data.

Until then, some hints about what to look for in the logfile:

4.2.1: Crank header and references

First, Crank outputs its header and a customized reference section to use when writing a paper. It then outputs the XML that was input into it, to allow for easier debugging of program operation.

4.2.2: Conversion to Scalepack files

Crank then proceeds to convert the input .mtz file into Scalepack .sca files for the various direct methods programs. Information from this is added to the logfile.

4.2.3: Extraction and Scaling

Crank then proceeds to extract all the columns it knows about from the input .mtz file. These column names are explicitly defined in the main Crank input XML file. All these extracted columns are then scaled with the CCP4 program SCALEIT. All logfile information from SCALEIT is appended to the Crank logfile. This is an important step to examine the logfile thoroughly, as there are some important pieces of information in here. Specifically, from the SCALEIT documentation: "It is important to look at the final analysis which compares <FP **2> and <FPH**2> after scaling. If their ratio is not near to unity, something has gone wrong!"

4.2.4: Superimpose substructure on model

If the free Emma program from the CCTBX project has been installed, it is possible to obtain the origin shift between the substructure coordinates obtained by one of the direct methods programs as compared to another set of coordinates. This allows for the comparison of phases at later steps of the structure determination pipeline. All output from the Emma program is appended to the logfile. This includes information about the atomic positions of reference and trial substructures, number of matching atoms and RMS between these atoms. In addition, it contains the transformed coordinates of the trial molecule, origin shifted to those of the model molecule.

4.2.5: Individual Program Logfiles

In addition to these standard Crank logfiles, depending on the programs that were invoked by Crank, additional sections will be appended to the logfile for each program that is run. In the next section we will talk about the program output of the crystallographic programs in Crank.

4.2.5.1: DREAR output

The DREAR suite outputs many useful logfiles as it runs. These files end with the suffix .lp and are in the crank/1-crunch2/mtz2drear directory. For more information, please visit the DREAR website. http://www.hwi.buffalo.edu/SnB/DREARhelp/DREAR.htm

4.2.5.2: CRUNCH2 output

For output, CRUNCH2 produces an atomic coordinates file for each trial specified by the user. These files have a simple four-column format, with the first column specifying the relative occupancy of the site, and the next three columns specifying the X, Y and Z coordinates of this atom. This format is shown below:

15.0          0.3040    0.0730    0.3246
13.8          0.6066    0.0658    0.5039
11.6          0.7966    0.0535    0.2389
10.8          0.6472    0.0615    0.5083
0.1621

In addition, CRUNCH2 has three main output files, "short.out", which is included in the Crank output file, "output", which contains much more detailed output information from the CRUNCH2 run, and "hits", which exhaustively details the progress of each trial run. It is often of use to examine the "output" file in cases of problems.

4.2.5.3: SHELXC/D output

For all information about SHELXC/D/E output, please visit the SHELX website at:


http://shelx.uni-ac.gwdg.de/SHELX/

4.2.5.4: BP3 output

BP3 produces a single unified log file which is included in the Crank output. First, BP3 prints it's header, followed by it's interpretation of the contents of the command file. In case of problems, it is advised to make sure the command file and the commands interpreted by BP3 are as similar as possible. In default mode, BP3 then proceeds with two refinement cycles, one just refining occupancies and one refining all parameters.

After refinement of the parameters, BP3 then prints out the coordinates, occupancies and B-factors of the input atoms and outputs its best phase estimates and phase probability statistics to the output MTZ file.

4.2.5.5: DM output

When running with SAD or SIR(AS) experiements, we do not have knowledge of which enantiomorph is the correct one, so we run both, one as "hand1" and the enantiomorph as "hand2". The DM output from these two hands is appended to the Crank logfile. Important pieces of information are if in the sections displayed through the macromolecule are contiguous.

For more information, please consult the DM manual directly.

4.2.5.6: SHELXE output

For all information about SHELXC/D/E output, please visit the SHELX website at:

http://shelx.uni-ac.gwdg.de/SHELX/

Chapter 5. Chapter 5: What to do if you encounter problems

Table of Contents

5.1: Problems with the Crank GUI
5.2: Problems with the Crank engine
5.3: Problems with CRUNCH2
5.4: Problems with BP3

5.1: Problems with the Crank GUI

5.2: Problems with the Crank engine

5.3: Problems with CRUNCH2

5.4: Problems with BP3

Chapter 6. Chapter 6: Crank Internals

Table of Contents

6.1: Crank XML
6.2: Crank directory structure
6.3: Crank Tcl / Crank CCP4i

6.1: Crank XML

Crank uses XML for communication both inside and outside the suite. Inside the suite, Crank uses XML to construct command files for the various programs in Crank. It also uses XML to transmit coordinate data, program output information, and timing results.

XML (eXtensible Markup Language) is very similar to the HTML language (Hypertext Markup Language) used to construct web pages. Like HTML it uses tags to annotate, or mark-up text files, adding semantic knowledge to the flat text file. An example of XML is shown below:

<dataset_info>
<num_atoms>3</num_atoms>
</dataset_info>

In the XML shown above, you can see some essential features of XML. First, XML uses tags to mark up the document. These tags are enclosed in angle braces "<" and ">". Second, you can see that there is always a start tag (beginning with "<" and a matching end tag, staring with "</".

Crank uses this structured markup language to facilitate the passing of information from one crystallographic program to another.

6.2: Crank directory structure

Crank uses a hierarcherically structured directory system for its run directory, in order to maintain order with all the different runs of the various crystallographic programs. There are two types of directories under the Crank directory hierarchy, directories where crystallographic programs are run, and directories where information is collected.

The directories where programs are run in always start with a number and a dash as in "1-crunch2" or "2-bp3". The first number signifies the order in which this program was run in the pipeline and the second identifier shows which program was run.

Inside these directories are where the various crystallographic programs are run. To simplify a complex situation, inside each of these directories are all the files produced by a run of the given crystallographic program. These directories are constructed by Crank, which first builds a shell script to run the program. This shell script is designed to be as close as possible to the example scripts generated by the program author. Then, Crank copies all the requisite data files for the program into the run directory.

Then, Crank simply executes the shell script that it has built, timing the run and collecting results. The results are then converted into XML format and the reflection file data is converted into MTZ format.

6.3: Crank Tcl / Crank CCP4i

Chapter 7. Chapter 7: More Information

Table of Contents

7.1: Crank
7.2: CRUNCH2
7.3: BP3
7.4: SHELX
7.5: CCP4/CCP4i

7.1: Crank

More information about Crank can be found both on the Crank website and in the Crank distribution. In addition, a paper discussing Crank and it's results has been published.

The Crank website can be found at http://www.bfsc.leidenuniv.nl/software/crank. On it, you can find lots of information about Crank, from how to run Crank to how to interpret results. There is also a SAD Tutorial on the Crank website that has detailed steps of how to perform a SAD experiment with Crank.

In the Crank distribution, there are documents, test directories and PDF versions of some of the papers published about Crank and some of the programs that comprise Crank. All documents are found in the "doc" directory of the Crank distribution, and include README's and .pdf versions of papers. The following papers have been included

wd5000r.pdf Navraj S. Pannu, Airlie J. McCoy and Randy J. Read Application of the complex multivariate normal distribution to crystallographic methods with insights into multiple isomorphous replacement phasing Acta Cryst. (2003). D59, 1801-1808

wd5005r.pdf Navraj S. Pannu and Randy J. Read The application of multivariate statistical techniques improves single-wavelength anomalous diffraction phasing. Acta Cryst. (2004). D60, 22-27

gc0004.pdf J.L van der Plas, R.A.G. de Graaff, and H. Schenk On the Use of Eigenvalues and Eigenvectors in the Phase Problem Acta Cryst. (1998). A54, 262-266

gc0005.pdf J.L. van der Plas, R.A.G. de Graaff, and H. Schenk Karle-Hauptman Matrices and Eigenvalues: a Practical Approach Acta Cryst. (1998). A54, 267-272

jn0094.pdf Rudolf A.G. de Graaff, Mark Hilge, Jaco L. van der Plas and Jan Pieter Abrahams Matrix methods for solving protein substructure of chlorine and sulfure from anomalous data Acta Cryst. (2001). D57, 1857-1862

7.2: CRUNCH2

There is more information about CRUNCH2 at the CRUNCH2 website:

http://www.bfsc.leidenuniv.nl/software/crunch2

In addition, there are three papers included in the "doc" directory in the Crank distribution that relate to Crunch2

gc0004.pdf J.L van der Plas, R.A.G. de Graaff, and H. Schenk On the Use of Eigenvalues and Eigenvectors in the Phase Problem Acta Cryst. (1998). A54, 262-266

gc0005.pdf J.L. van der Plas, R.A.G. de Graaff, and H. Schenk Karle-Hauptman Matrices and Eigenvalues: a Practical Approach Acta Cryst. (1998). A54, 267-272

7.3: BP3

There is more information about BP3 at the BP3 website:

http://www.bfsc.leidenuniv.nl/software/bp3

In addition, there are two papers included in the Crank distribution that refer to the algorithms used in BP3.

wd5005r.pdf Navraj S. Pannu and Randy J. Read The application of multivariate statistical techniques improves single-wavelength anomalous diffraction phasing. Acta Cryst. (2004). D60, 22-27

7.4: SHELX

Although Crank has an interface to SHELX, SHELX is not part of the Crank suite. To obtain the SHELX suite, and for more information, please go to the SHELX website:

http://shelx.uni-ac.gwdg.de/SHELX/

7.5: CCP4/CCP4i

Crank uses libraries from the CCP4 project extensively. In fact, for it's interface, Crank uses the CCP4i interface and libraries. In addition, Crank uses many programs from the CCP4 suite including SFTOOLS, TRUNCATE, WILSON, MTZ2VARIOUS, CAD, SFTOOLS, MTZMADMOD, F2MTZ and others.

In addition, for density modification, Crank uses the DM program which is included in the CCP4 distribution. For more information on DM, please consult the CCP4 program documentation or the following paper:

Cowtan, K. (1994). Joint CCP4 and ESF-EACBM Newsletter on Protein Crystallography, 31, p34-38.

For more information on the CCP4 please visit the CCP4 website:

http://www.ccp4.ac.uk/

Chapter 8. Chapter 8: References

Table of Contents

Bibliography

Bibliography

Steven R. Ness, Rudolf A.G. de Graaff, Jan Pieter Abrahams, and Navraj S. Pannu. “Crank: New methods for automated macromolecular crystal structure solution”. Structure. 12. 1753-1761. October 2004. .

J. P. Abrahams and A. G. W. Leslie. 1996. “Methods used in the structure determination of bovine mitochondrial F1 ATPase.”. Acta Cryst.. D52. 30-42. . .

R.H. Blessing and G.D. Smith. 1999. “Difference Structure Factor Normalization for Determining Heavy-Atom or Anomalous Scattering Substructures.”. J. Appl. Cryst.. 32. 664-670. .

J. S. Brunzelle, P. Shafaee, X. Yang, S. Weigand, Z. Ren, and W. F. Anderson. 2003. “Automated crystallographic system for high-throughput protein structure determination.”. Acta Cryst.. D59. 1138-1144. .

M.C. Burla, B. Carrozzini, G.L. Cascarano, C. Giacovazzo, and G. Polidori. 2003. “SAD or MAD phasing: location of the anomalous scatterers.”. Acta Cryst.. D59. 662-669. .

K. Cowtan. 1994. “Density Modification”. Joint CCP4 and ESF-EACBM Newsletter on Protein Crystallography. 31. p34-38. .

Z. Dauter, M. Dauter, E. de La Fortelle, G. Bricogne, and G. M. Sheldrick. 1999. .“Can anomalous signal of sulfur become a tool for solving protein crystal structures? ”. J. Mol. Biol.. 289. , 83-92. . .

Z. Dauter and D. A. Adamiak. 2001. “Anomalous signal of phosphorus used for phasing DNA oligomer: importance of data redundancy.”. Acta Cryst.. D57. , 990-995. . .

Z. Dauter, M. Dauter, and E. Dodson. 2002. “Jolly SAD.”. Acta. Cryst.. D58. 494-50. .

E. Dodson. 2003. “Is it jolly SAD? ”. Acta. Cryst.. D59. , 1958-1965. .

P. Emsley. 1999. “Datasets to Maps: A Wrapper for SHELXS and the CCP4 Program Suite”. CCP4 Newsletter on Protein Crystallography. 36. .

R. A. G. de Graaff, M. Hilge, J. L. van der Plas, and J. P. Abrahams. 2001. “Matrix methods for solving protein substructures of chlorine and sulfur from anomalous data.”. Acta Cryst.. D57. 1857-1862. .

R.W. Grosse-Kunstleve and P.D. Adams. 2003. “Substructure search procedures for macromolecular structures.”. Acta. Cryst.. D59. 1966-1973. .

T. Hamelryck and M. Kjeldgaard. 2001. “An Open Source Multi-purpose Programming Environment for Macromolecular Crystallography.”. CCP4 Newsletter on Protein Crystallography. 39. .

N. Ito, K. Kobayashi, H. Sakamoto, and H. Nakamura. 2002. “Development of PDBj-ML.”. Acta Cryst. A58 Supplement. C367. .

J. Karle and H. Hauptman. 1950. “The phases and magnitudes of the structure factors.”. Acta Cryst.. 3. 181-187. .

E. de la Fortelle and G. and Bricogne. 1997. Methods Enzymol.. 276. 472-494. . .

A. G. W. Leslie, H. R. Powell, G. Winter, O. Svensson, D. Spruce, S. McSweeney, D. Love, S. Kinder, E. Duke, and C. Nave. 2002. “Automation of the collection and processing of X-ray diffraction data - a generic approach.”. Acta Cryst.. D58. 1924-1928. .

R.J. Morris, A. Perrakis, and V.S. Lamzin. 2002. “ARP/wARP's model-building algorithms. I. The main chain.”. Acta Cryst.. D58. 968-975. .

G.N. Murshudov, A.A. Vagin, and E.J. Dodson. 1997. “Refinement of Macromolecular Structures by the Maximum-Likelihood Method.”. Acta Cryst.. D53. 240-255. .

N. S. Pannu, A. J. McCoy, and R. J. Read. 2003. “Application of the complex multivariate normal distribution to crystallographic methods with insights into multiple isomorphous replacement phasing.”. Acta Cryst.. D59. 1801-1808. .

N. S. Pannu and R. J. Read. 2004. “The application of multivariate statistical techniquest improves single-wavelength anomalous diffraction phasing.”. Acta Cryst.. D60. 22-27. .

A. Perrakis, R. Morris, and V.S. Lamzin. 1999. “Automated protein model building combined with iterative structure refinement.”. Nature Struct. Biol. . 6. 458-463. .

J. L. van der Plas, R. A. G. de Graaff, and H. Schenk. 1998. “On the Use of Eigenvalues and Eigenvectors in the Phase Problem.”. Acta Cryst.. A54. 262-266. .

J. L. van der Plas, R. A. G. de Graaff, and H. Schenk. 1998. “Karle-Hauptman Matrices and Eigenvalues: a Practical Approach.”. Acta Cryst.. A54. 267-272. .

E. Potterton, P. Briggs, M. Turkenburg, and E. Dodson. 2003. “A graphical user interface to the CCP4 program suite.”. Acta Cryst.. D59. 1131-1137. .

T.R. Schneider and G.M. Sheldrick. 2002. “Substructure solution with SHELXD.”. Acta Cryst.. D58. 1772-1779. .

G.D. Smith. 2002. “Matching selenium-atom peak positions with a different hand or origin.”. J. Appl. Cryst.. 35. 368-370. .

T.C. Terwilliger and J. Berendzen. 1999. “Automated MAD and MIR structure solution.”. Acta Cryst.. D55. 849-861. .

C.M. Weeks and R. Miller. 1999. “The design and implementation of SnB v2.0”. J. Appl. Cryst. 32. 120-124. .

M. D. Winn, A. W. Ashton, P. J. Briggs, C. C. Ballard, and P. Patel. 2002. “Ongoing developments in CCP4 for high-throughput structure determination.”. Acta Cryst.. D58. 1929-1936.. .