================================================== CCP4 Program suite Frequently asked Questions list ================================================== Maintained by Paul Hempstead ***************************************************************** This list is based on OFAQ/rfaq from Gerard and Morten: gerard@xray.bmc.uu.se morten@oase.kemi.aau.dk ***************************************************************** Started: 950228 Version = 0.1 @ 950309 (17 questions & answers) Version = 0.2 @ 951905 (58 questions & answers) This version = 0.3 @ 953110 (57 questions & answers) Mail contributions to: p.hempstead@sheffield.ac.uk ***************************************************************** Format: o actual faq list delimited by a line containing -START- and one containing -END- o each question delimited by Q dot NNN, where NNN is a unique number, at the start and a line containing 7 - signs at the end for each question: Q dot NNN + one line describing question S dot NNN + one or two keywords A dot NNN + any number of lines for answer Count the number of questions as follows: grep '^Q\.' ccp4.faq | wc -l ***************************************************************** Formatting information is contained in this file for use in preparing the html pages: A blank line causes a new paragraph to begin in the html file. Any block of lines starting with a space are treated as special: Lines starting space o, space *, or space - are treated as itemised lists ***************************************************************** Introduced new keywords, one per question. Used so far: Keywords: Amore Amore_E Amore_input Amore_solutions Analysis Anomolous_data Archive Averaging_programs Creating_MTZ DANO_definition DM_map Density_modification Equivalent_reflections FFT_problem Free_R Hydrogen_positions IRIX6 MIR MR MTZ_conversion MTZ_manipulation Mama_mask Map_conversion Mask_generation Mask_manipulation Mosflm MR_and_DM NPO_input PLOT84_conversion Plotting_program References Reindexing Scala Scaling Sfall_documentation T0/0_function XDS address binaries ccp4bb contributions documentation documents ftp_problems general index information installation introduction memory_allocation newsletter runtime software source_code To get an up-to-date list, execute the following C-shell script: #!/bin/csh -f set x=`grep '^S\.' ccp4.faq | cut -c9-80` if (-e q.$$) \rm q.$$ touch q.$$ foreach y ($x) echo $y >> q.$$ end set x=`cat q.$$ | sort | uniq` echo "Keywords:" echo $x \rm q.$$ exit 0 To get the complete list of questions for a table of contents: grep ^Q\. ccp4.faq ***************************************************************** (I) INTRODUCTION (II) GENERAL_PROBLEMS (III) DATA_PROCESSING AND SCALING (IV) MIR_AND_MOLECULAR_REPLACEMENT (V) REFINEMENT (VI) PHASE_IMPROVEMENT (VII) ANALYSIS (VIII) PRESENTATION_OF_RESULTS (IX) MISCELLANEOUS (X) OPERATING SYSTEM PROBLEMS ***************************************************************** ---START--- (I) INTRODUCTION Q.001 - Why does this FAQ exist? S.001 - introduction A.001 - This FAQ contains answers to questions regarding the CCP4 program suite for protein crystallography. If you have any comments about this FAQ, please contact p.hempstead@sheffield.ac.uk Any contributions to this FAQ will be gratefully received. So, if you have had questions answered by CCP4 and you think other people would benefit from the information please e-mail the Q&A to p.hempstead@sheffield.ac.uk This FAQ will be updated (and expanded) regularly. The current version is held at: ------- Q.002 - What is CCP4? S.002 - introduction A.002 - CCP4 is the Collaborative Computational Project no. 4 in protein crystallography. It is one of the older CCPs, set up under (originally) a UK Science Research Council initiative to encourage collaborative computational work, between UK groups, . CCP4 is now funded by the UK Biotechnology and Biological Sciences Research Council and administered from Daresbury Laboratory (a UK national lab) . The CCP4 suite of programs is developed by working crystallographers and distributed in source form (on purpose). It is intended to be a resource for crystallographers worldwide, which we hope people will use and contribute to. ------- Q.003 - How do I contact CCP4? S.003 - address A.003 - Presumably you already have! We prefer contact by e-mail, if possible: Secretary to CCP4 | Telephone: (+44) 1925 603530 (direct line) Daresbury Laboratory | Facsimile: (+44) 1925 603124 Warrington WA4 4AD | e-mail: ccp4@dl.ac.uk UK | A bulletin board service for the suite is also available (see Q.008). ------- Q.004 - How do I obtain the CCP4 software? S.004 - software A.004 - The best way is by Internet anonymous ftp to URL . If necessary we can supply it on tapes (for a media/handling charge) and in principle it is also available by e-mail (send `HELP' to info-server@dl.ac.uk) but it's really too large for that. See Q.017 about size. The software is available under the terms of the licence agreement and copying terms, copies of which are on the ftp server and at the back of the printed manual (or we can snail one too you). There is no charge for not-for-profit organisations, but for-profit ones are asked to pay. In the latter case please contact the Secretary (see Q.003) for current arrangements. Note that CCP4 is not properly `free' in the GNU General Public Licence or other usual interpretations of `free software'; in particular, it is not freely redistributable (see the copying terms). You may regard this as a bug. ------- Q.005 - How do I get binaries? S.005 - binaries A.005 - We don't distribute binaries. A fee might persuade us to make them for you for a system to which we have access. We want you to have the source (and fix it!) and making and distributing binaries is usually too problematic for various reasons. ------- Q.006 - How do I get copies of the `Study Weekend' proceedings? S.006 - documents A.006 - This is a list of the available CCP4 Daresbury Study Weekend Proceedings. Due to the large demand for copies of the proceedings we have had to start charging. Charges are intended to cover copying and distribution costs. Current charges are 10 UK pounds for each copy. In order to simplify things at this end please send a cheque for the approximate amount in your currency with your request. Cheques should be made payable to 'Daresbury Laboratory'. The charges may be waived under circumstances of hardship. No charges will be made for copies collected from Daresbury Laboratory. Please address requests to: The Librarian Daresbury Laboratory Warrington WA4 4AD, UK Phone: (+44/0)1925 603397, FAX: (+44/0)1925 603100 E-mail: library@dl.ac.uk 1980) Refinement of protein structures. DL/SCI/R16 1985) Molecular replacement. DL/SCI/R23 1987) Computational aspects of protein crystal data analysis. DL/SCI/R25 1988) Improving protein phases. DL/SCI/R26 1989) Molecular simulation and protein crystallography. DL/SCI/R27 1990) Accuracy and reliability of macromolecular crystal structures. 1991) Isomorphous replacement and anomalous scattering. DL/SCI/R32 1992) Molecular replacement. DL/SCI/R33 1993) Data Collection and Processing. DL/SCI/R34 1994) From First Map to Final Model. DL/SCI/R35 1995) Making the Most of Your Model. DL-CONF-95-001 1996) Macromolecular Reinement. DL-CONF-96-001 1997) Recent Advances in Phasing. DL-CONF-97-001 ------- Q.007 - How do I get the CCP4/ESF newsletter and study weekend announcements? S.007 - newsletter A.007 - Our snail-mail list is maintained by the Daresbury library (see Q.006.). Ask to be put on the CCP4 mailing list. Recent copies of the newsletter are available at . ------- Q.008 - How do I get on/off the CCP4 `bulletin board' mailing list? S.008 - ccp4bb A.008 - This is supposed to be automatic, but there are often problems. If the following messages: sub ccp4bb or unsub ccp4bb respectively, sent to `ccp4bb-requests@dl.ac.uk' don't work, ask `ccp4bb-request@dl.ac.uk' to do it manually. Please try to unsub *before* your e-mail address changes. ------- Q.009 - Where can I find other relevant software/information? S.009 - information A.009 - The crystallography section of the World Wide Web virtual library is at. Also, perhaps ask on usenet group bionet.xtallography; your news admin may have to take special action to get bionet. ------- Q.010 - (How) can I contribute (software) to CCP4? S.010 - contributions A.010 - Please offer us any programs in your gift which you think fill gaps in CCP4 and send any code fixes, useful examples, documentation re-writes etc. It is as well to check with us first before expending much effort on something it may just be in hand already. If you're offering programs, it's useful if they already use our library routines for MTZ and map handling etc., and have conformant `keyworded' input. The manual has a (currently deficient) section on writing code. ------- Q.011 - Is there a program to do X? S.011 - index A.011 - Have you tried using `apropos'/`man -k' under Unix and looked in the main and program index in the printed manual? ------- Q.012 - Why isn't your documentation in format X? S.012 - documentation A.012 - The online program documentation is now mostly in HTML format. The manual however is in LaTeX, because that's the only reasonable thing to do. If you wish to write the technology to convert LaTeX to efficient and portable HTML without losing information -- consider the math and cross- referencing macros -- please do so and send it to us and the grateful world. Yes, hypertext is a very good idea and a great deal of work to do properly; no HTML is not identical with hypertext and is (at least currently) deficient in various ways for producing scientific documentation. ------- Q.013 - Where can I find newly corrected versions of programs? S.013 - source_code A.013 - Where bugs have been found (and fixed) in programs, the corrected versions may be made available in a prerelease directory at the anonymous Daresbury ftp address: . ------- Q.014 - Where are the old ccp4bb messages? S.014 - Archive A.014 - No archive of these questions is available. However, this FAQ has been set up to try to keep a record of questions and answers from ccp4bb [note that the ccp4 address is often a more appropriate one for enquiries]. This FAQ will be made available as an ascii file and provided with a searching script (for UNIX and VMS) along the lines of the O FAQ (from which much has been borrowed). ------- Q.015 - What is the definition of DANO? S.015 - DANO_definition A.015 - Dano, the anomolous difference, and the associated F's are defined as: Dano = (Fhkl - F-h-k-l) Fmean = (Fhkl + F-h-k-l)/2 Fhkl = Fmean + Dano/2 F-h-k-l = Fmean - Dano/2 ------- Q.016 - How do I reference CCP4? S.016 - References A.016 - Please cite the following paper: Collaborative Computational Project, Number 4 (1994) Acta Cryst D50 760-763. `The CCP4 Suite: Programs for Protein Crystallography' In addition, users are asked to reference individual programs where use of these programs played a key role in their investigations. Your cooperation is requested, as use of this new reference will allow us to collate citations of CCP4. ------- Q.017 - Size of Suite and recommended minimum installation. S.017 - Software A.017 - The whole suite is quite large and can be too much for some machines. For instance all the files (including tar files) on "/pub/ccp4/" are in total approximately 70MB. The untared files are 90MB worth, 50MB for the Laue Suite (only the compressed form is present in the ftp area) and 40MB for the rest. Additional space is needed for scratch files and binaries, this will be machine dependent. We recommend that the directories and files needed for a minimum installation are: pub/ccp4/ccp4/* pub/ccp4/ccp4/lib/src/* pub/ccp4/ccp4/lib/data/* pub/ccp4/ccp4/src/* and sub-directories pub/ccp4/ccp4/include/* pub/ccp4/ccp4/lib/etc/ not essential, PROCHECK ones useful pub/ccp4/prerelease maybe not all files needed Further you may well consider that the documentation is essential as well. Note that the /man and /doc files are mostly redundant. Not all programs have man pages though. The whole "/pub/ccp4/ccp4/" directory and sub-directories, the core of the Suite, are about 40MB (~27MB excluding compressed tar files). Obviously, in the minimum installation not all programs mentioned in the manual will be available. For information on how to obtain the Suite see Q.004. ------- ***************************************************************** (II) GENERAL_PROBLEMS Q.101 - Why do I get a `write error' when building CCP4? S.101 - installation A.101 - You have insufficient space in the disc partition where you're doing the building, insufficient quota or insufficient space in the compiler's temporary directory. These are local problems. The latter may be solvable on Unix by telling the compiler to use a different place or to pipe between stages. ------- Q.102 - Why do I get `cannot mix Elf and COFF objects.' when building CCP4? S.102 - installation A.102 - You are probably trying to link object files created under a previous version of the operating system which used a different format. This is a frequent complaint from people building CCP4 on Irix5 somewhere they have used under Irix4 and not cleaned up. Do `make realclean' in the `ccp4' directory and start again with configure and make. ------- Q.103 - Why do I get an `IOT trap' error? S.103 - runtime A.103 - Presumably you're running under Irix. This probably indicates a floating point error. The default ccp4.setup sets the environment variable TRAP_FPE to abort on overflow, divide-by-zero and invalid operands. See handle_sigfpes (3F) for more information. You may be able to proceed by unsetting TRAP_FPE, but that may give give you wrong or silly results. If you are capable of finding the problem yourself with your data using a debugger it is helpful -- please send us the information. Otherwise to fix it we probably need your input files to run here; ask about uploading them. ------- Q.104 - Why do I get `Open failed:' error? S.104 - runtime A.104 - If there isn't a helpful associated message from the system library it is a bug, possibly in the system, library, possibly in the CCP4 library -- please report it. If the complaint is about a `new' file existing then you have probably got the logical name CCP4_OPEN set to `NEW'. If you want to overwrite files that already exist, change the value of the logical name CCP4_OPEN to `UNKNOWN' by any of the means explained in the manual. This can be done locally to a program or script if necessary, e.g. if running cycles of a program. Another thing to check is that the environment variable $CCP4_SCR has been defined properly and that this directory exists. Scratch files are opened by many programmes which are not necessarily documented. The best thing to do, if possible, is to examine an old log file to see which file was not opened. ------- Q.105 - Why does program X not produce the right output? S.105 - general A.105 - In most cases there will be an error in the input file, which may be very difficult to spot. If you have any specific examples of jobs which don't run successfully then you should contact ccp4@dl.ac.uk specifying the nature of the error, perhaps including your input control file and your output log file. It is important that users inform ccp4 of problems so that corrective action can be taken for the benefit of the user community - in the case of coding errors corrections can be made - in the case of user errors additions can be made to the documentation to help others. ------- Q.107 - How do I intercovert my XPLOR/CCP4/FRODO/O maps? S.107 - Map_conversion A.107 - The best(?) solution is to use mapman (from Uppsala) for which you need to contact Gerard Kleywegt, G.Kleywegt@xray.bmc.uu.se, or see www page . This allows the interconversion of several map formats from and to CCP4. It may be easiest to exchange reflexion files between programs (see `f2mtz' and `mtz2various') and just do the Fourier transforms. ------- Q.108 - Atoms being ignored in SFALL? S.108 - Sfall_documentation A.108 - Some users have asked why SFALL fails to calculate structure factors for a PDB file which contains atoms other than O, C, N & S. The answer is that these formfactors have been hard-wired into the program (they are always present) and any other atoms need to be specified in the input file. e.g. for a structure containing P use: FORM NGAUSS 5 P Another problem that can catch people out is that the PDB format requires atom names to consist of the chemical symbol (right justified) in columns 13 and 14, a remoteness indicator in column 15 and a branch designator in column 16. If you have a coordinate file in which the atom chemical symbol has not been right justified you may end up with your atoms being misunderstood: e.g. the first three atoms would be read as C, Cu, and Cl, but the fourth atom would be read as a carbon atom NOT as a chlorine. ATOM 1 CA ATOM 1 CU ATOM 1 CL ATOM 1 CL For more on the PDB file standards, retrieve postscript file pub/format.desc.ps from pdb.pdb.bnl.gov. ------- Q.109 - SFALL problem transforming map? S.109 - FFT_problem A.109 - Concerning the `sfall' error "WARNING - your map spacegroup is different to the program default one " "Fatal disagreement between input info and map header". Consult the `sfall' documentation for MODE SFCALC and NOTES ON MAPIN. Points to remember: * ALWAYS use the same FFTSPG and SFSG Use the defaults for everything else - in particular: * NEVER use the `fft' AXIS keyword - the fft and sfall defaults are the same for the same FFTSPG and SFSG, and you may well make a mistake.. * Dont try to set GRID unless you want a finer one than the default. * NEVER use the `fft' XYZLIM - the program will find the default which will agree with sfall. * NEVER use the SYMM keyword - this will be read from the mtz file. ------- Q.110 - Which reflections are equivalent in my spacegroup? S.110 - Equivalent_reflections A.110 - A question was posed as to which reflections are equivalent to (6,4,37) in spacegroup P622 (177). The answer is included with the general method for calculating equivalent reflections in any spacegroup: The equivalent positions in the ccp4 symmetry library (and the international tables as well) are given for real space. However symmetry relations between reflections are in reciprocal space. To get the reciprocal space symmetry from the real space one you have to take the transpose of the symmetry matrix. For example: -y, x-y, z is a real space symmetry operation for P622. In matrix notation this is ( 0 -1 0) ( 1 -1 0) ( 0 0 1) The reciprocal symmetry operator now becomes ( 0 1 0) (-1 -1 0) ( 0 0 1) This corresponds to k, -h-k, l or (4,-10,37) for the (6,4,37) reflection. The other equivalent reflection can be calculated from the other symmetry operators. ------- Q.111 - How do I manipulate a CCP4 format mask? S.111 - Mask_manipulation A.111 - Try NCSMASK and MAPMASK. ------- Q.112 - How do I convert data files from format X to CCP4? S.112 - MTZ_conversion A.112 - There are two types of MTZ file, standard and multi-record. A standard MTZ file has all the information for a particular reflection in one record i.e. each hkl appears only once in the file. Multi-record files contain data which have not been merged, with potentially several records for each hkl. Standard MTZ files are produced by `f2mtz'; use this for importing data which don't require scaling and merging with CCP4 programs. Be careful to use all real number formats if you use `f2mtz's FORMAT option -- see the documentation. CAD will sort and move the reflections into the standard part of reciprocal space. `rotaprep' produces multi-record output suitable for scaling/merging with `scala' and/or `agrovata'. It can usually easily be extended for new input formats if necessary. ------- Q.113 - I am having problems with anonymous ftp! S.113 - ftp_problems A.113 - This is a list of some of the problems you might incur with our anonymous ftp server. Except where noted, the only solution is to try again at some other time and if a problem at our end persists, let us know (see Q.003). 1. Failure to connect at all means either the network or the server's host is down. 2. It isn't currently possible to give more helpful information when the server tells you it's shut down (e.g. for updating the distribution). If it's persistently shut down, report it. 3. The error "can't set guest privileges" typically indicates a problem with the server's disc or local network access. Report it if persistent. 4. If your login attempt is otherwise rejected it should say why. If not, it's a bug. If the maximum number of connections is continually exceeded it may be because old ones haven't timed out correctly; we can kill them if necessary. Messages about you not having a valid nameserver require action at your end; either move to a different system or get yours registered in the `DNS' (domain name service); this will take time to propagate to us anyhow. You can check your DNS entry with nslookup `hostname` 5. The connection is very slow/unreliable. This can be a major problem from the US since the trans-Atlantic link is currently inadequate; other national networks may have their own problems. The `traceroute' program can reveal the hiatus. Try again at an off peak time. Our network experts say that the best time for UK/US network bandwidth is usually early in our morning, Sunday especially(!). One can use `ftpcat' with `cron' or similar to avoid losing the sleep. If a slow ftp session fails when you've got part of the file, try the `reget' command which our server supports. In the canonical ftp client, `reget file' is like `get file' except that it appends to what you already have locally. There may in future be a mirror site in the US. Details will be given if/when this is created. ------- (III) DATA_PROCESSING AND SCALING Q.201 - How can I manipulate my MTZ data file to remove Fobs=0.0? S.201 - MTZ_manipulation A.201 - MTZUTILS will rearrange, combine, and select data from a MTZ reflection file. CAD will Combine Assorted Data files (and expand them to lower symmetry, sort the data, etc). However, neither of these will allow you to exclude reflections with Fobs=0.0. You could, of course, convert your data file to ascii, using MTZ2VARIOUS, and use dataman (from Uppsala) to edit your data and then convert back to the MTZ format with F2MTZ. Another option is to use SFTOOLS available from: ------- Q.202 - Is it possible to carry a phase information through SFALL? S.202 - MTZ_manipulation A.202 - The answer to the general question of whether it is possible to carry information (in an MTZ file) through a program passively is Yes. To carry all the columns from the input MTZ file through SFALL specify the following in your SFALL job: LABOUT ALLIN ------- Q.203 - How do I convert a Denzo file to MTZ format? S.203 - Creating_MTZ A.203 - See Q.112 and note the `rotaprep' documentation (especially concerning `scalepack' and `denzo' output options). ------- Q.205 - Mosflm and LAUE queries S.205 - Mosflm A.205 - The LAUE and "mosflm" systems are associated with CCP4 (in particular, using the CCP4 library) but aren't part of it, and questions may be best addressed elsewhere. The distributions are at and (Unix and VMS distributions). Questions about LAUE are probably best directed to the maintainer, John Campbell . Discussion about "mosflm" does occur on the "ccp4bb" list and you may get helpful information from that, but the maintainer and ultimate expert is Andrew Leslie . ------- Q.206 - Scala/agrovata aggravation. S.206 - Scala A.206 - In response to: Probably due to a lack of patience today, this standard deviation fudge in agrovata/scala is driving me crazy. Suppose you put in the standard option: sdfac 1.0 sdadd 0.0 and out comes the standard deviation analysis of agrovata, with overall figures of 0.16 (mean) and 1.74 (sigs). What values would you then use for sdfac and sdadd, and how do you get at them, and when should you start worrying that the data you're scaling were turned to garbage a couple of steps earlier? Answer, the idea is to try different values of SDFAC and SDADD to get the mean to approximately 0 and sigma about 1. Generally, you need to try 10-20 combinations to get very good values. It is quite common to have SDFAC of 1.2 to 1.3 and SDADD can be 0.02 or 0.03. Read mosflm_user_guide.doc section 5.3 on how to access data quality... well I enclose it here: 5.3 Checking the quality of the data Although there are indicators of data quality in MOSFLM (in particular the I/sig(I) and Rsym values as a function of resolution), the only satisfactory way of assessing data quality is to look at the results of merging measurements of symmetry related reflections using AGROVATA. Remember that the R-factor alone is not a good indicator, it will always be high for weak data. What is probably more important is the standard deviation analysis at the end of AGROVATA. If this suggests that the observed agreement is that expected based on the standard deviations (ie SIGM is 1.0) then you cannot hope to do any better. Inevitably there are errors which are not accounted for in the estimated standard deviation. Thus it is quite normal to have to boost the standard deviations by 20-30% (ie an SDFAC of 1.2-1.3) to achieve a SIGM value of 1.0. In addition, the agreement is generally worse for the strongest data, so an SDADD of between 0.02 and 0.03 is quite common. If these parameters have to be made significantly greater than this to achieve a SIGM value of 1.0 across the intensity range, then this indicates problems with the processing which should be investigated. One possibility is the presence of a few large outliers, which can destroy the SIGM analysis...look at the monitored reflections for evidence of this. In cases where the crystal has a high mosaic spread, pay particular attention to the partial bias analysis. If this is more than 1-3%, then the mosaic spread has probably been incorrectly defined. In difficult cases it may well be necessary to process a dataset several times (eg with different mosaic spreads, or different numbers of images included in post refinement or in forming the standard profiles) in order to achieve the best final dataset, but this should be simply a case of using more cpu time, and not require a lot of intervention. ------- Q.207 - How do I reindex my data? S.207 - Reindexing A.207 - The program REINDEX will allow you to reindex your data, and will keep track of your anomolous data for you. ------- Q.209 - Why doesn't my data scale very well? S.209 - Scaling A.209 - There are 3 possible problems: 1) the crystals are different: this is by no means unlikely 2) the scaling has not properly converged. This is quite likely if your crystals are on very different scales, or cover different resolution ranges. 3) the datasets are not indexed in the same way. We have had several cases recently in trigonal and hexagonal space groups (and even one in monoclinic) where the indexing is perfectly correct on its own but is inconsistent with a second crystal owing to the ambiguity in the definition of the axes in these space groups. Yes: the program REINDEX will allow you to reindex data. ------- Q.210 - Help with XDS data processing S.210 - XDS A.210 - See also Q.112 and the MAR documentation for their version of XDS. To process XDS.HKL data (scaled, unmerged), use `rotaprep' and `agrovata'. To process UNIQUE.HKL or marscale output, `f2mtz' followed by `truncate' is appropriate, or `rotaprep' if you want to merge the data with others. ------- (IV) MIR_AND_MOLECULAR_REPLACEMENT Q.301 - Why does AMORE run out of space? S.301 - memory_allocation A.301 - See the documentation (`Memory allocation') for a discussion of amore's memory allocation and how to change it. ------- Q.302 - How do I incorporate E's in AMORE S.302 - Amore_E A.302 - See the script examples/unix/runnable/RF-with-Es for a template for running a rotation function with Es rather than Fs. The CCP4 version of the Amore translation function step cannot read structure factors from `ecalc' so you will either have to use Fs for the translation function with Amore or use `tffc' for the TF step (see examples/ (since you would otherwise be comparing E's with F's)). You can use sharpened Fs. ------- Q.303 - Using anomolous scattering to determine handedness. S.303 - Anomolous_data A.303 - In response to: I would appreciate it anybody could provide me literature references about the use of anomalous heavy atom data in the determination of handedness in the MIR method. (eventually based on considerations of the point group in the case of enantiomorphic space groups). Answer, see T.L.Blundell and L.N.Johnson (1976) Protein Crystallography 373-375. ------- Q.304 - Amore doesn't recognise my metal atoms. S.304 - Amore_input A.304 - See Q.108 regarding justifaction of atom names in the coordinate file. ------- Q.305 - How do I progress from MIR phases and a model? S.305 - MIR A.305 - In response to the following questions: - What is the best way to combine phases derived from MIR and those from my model? -How many amino acids (poly ALA) of my protein are necessary to give enough information - 50%, 75% or the most I have? - At which stage should I start to refine my poly-ALA model with XPLOR? And what protocol should I use? - Should I try to build as much backbone as possible into my density or should I already build side chains? - How important are errors that I have built into my model? - Any good literature references? Answer, one useful reference is F.M.D. Vellieux et al., Acta. Cryst. B46, 806-823. The problem of combining phase information is best dealt with using the program SIGMAA. ------- Q.306 - I need an introduction to molecular replacement. S.306 - MR A.306 - A number of texts contain very good introductions to MR: "The Molecular Replacement Method" M.G.Rossmann - editor, Gordon and Breach, Science Publishers, New York, N.Y., 1972. "The CCP4 Suite - Overview and manual" and references therein. "Proceedings of the CCP4 Study Weekend on Molecular Replacement", January 1992 available from Daresbury. "{AMoRe}: an Automated Package for Molecular Replacement", J. Navaza (1994), Acta Cryst., A50, 157-163. ------- Q.307 - Using the TO/O function S.307 - TO/O_function A.307 - In response to: I am trying to use the Harada et al., TO/O function as implemented in TFFC for a centered cell (spacegroup C2) and had a few questions: a) does one need to generate independent sets of structure factors and phases (using SFALL) for EACH of the crystallographic positions in the unit cell and do centering operations count as independent positions (i.e. should there be 2 or 4 positions in a C2 lattice)? Also, are the positions generated for a P1 cell or for the correct point group? b) for non-crystallographic symmetry it seems that the position of the first molecule needs to be defined; can this be done by going through the first run WITHOUT defining non-crystallographic symmetry and using the solution for all subsequent cases (the documentation seems to ontradict on this point). Answer, 1. You may wish to use the TO/O option. However, it involves significantly more CPU and seems not to give any better results than the standard T2 function. 2. The effect of centring is merely to multiply all structure factors by a constant (2 in the case of C2) which is taken care of in the scaling. In real space the molecules related by centring are always related by the same translation vector, so once you fix one you automatically fix all. 3. You have to run SFALL once for each molecule in the asymmetric unit, e.g. if you have 4-fold NCS you run SFALL 4 times, once for each of the RF solutions. Then you have to use a special option in the program CAD which collects together all the structure factor files and produces an output MTZ file which contains 1 pair of amplitude/phase columns for each molecule in the PRIMITIVE unit cell, e.g. for 4-fold NCS in C2 there would be 8 amplitude/phase pairs. 4. You run TFFC first for each independent molecule separately. Then fix the molecule which gave the best peak and search for the next best using the PART option. The solution for this 2nd molecule should be related by an origin shift to the initial solution you got for this molecule. Then fix molecules 1 & 2 and find the 3rd one, and so on. See the example in $CDOC/cad.doc . ------- Q.308 - How do I apply the MR solution from AMORE? S.308 - Amore_solutions A.308 - A number of people have had problems applying MR solutions from AMORE to produce the coordinates for the solution. Here is a stepwise guide to the process: In the begining: $! rotate and shift coordinates $! $AMORE XYZIN1 monomer2.pdb XYZOUT1 searchrot.pdb TABLE1 search.tab TITLE Produce Table for Search model VERBOSE TABFUN CRYSTAL ORTH 1 CELL_CRYS 81.381 81.381 31.580 90. 90. 120. MODEL 1 BREPLACE 0 BADD 0 SAMPLE 1 RESO 4 SHANN 2.5 END Run cross rotation, then translation: $amore TABLE1 search.tab HKLPCK0 e40tpck.mtz MAPOUT amore_trans1.map TRAFUN CB NMOL 1 RESO 10.0 4.0 PKLIM 0.5 NPIC 20 SYMM P3121 VERBOSE TITLE Translation Function of E40T in P3121 CRYSTAL ORTH 1 FLIM 0.E0 1.E8 SHARP 0.0 SOLUTIONRC 1 44.00 39.00 26.00 END Then the fit: $amore TABLE1 search.tab HKLPCK0 e40tpck.mtz FITFUN fitfun.txt FITFUN NMOL 1 RESO 10 4 ITER 10 CONV 1.E-3 TITLE E40T in P3121 Fitting Function VERBOSE SYMM P3121 CRYSTAL ORTH 1 FLIMI 0.E0 1.E8 SHARP 0.0 REFSOL AL BE Ga X Y Z BF SOLUT_1 1 44.00 39.00 26.00 0.4517 0.5916 0.1753 41.8 45.1 1 END Giving this: SOLUTIONF 1 40.18 39.98 28.06 0.45202 0.58890 0.18079 56.2 39.1 Now run PDBSET on the ****SEARCHROT.PDB**** with the correct (real) lattice parameters in the first line: $PDBSET XYZIN SEARCHROT.PDB XYZOUT FINAL.PDB ROTATE EULER 40.18 39.98 28.06 SHIFT FRACT 0.45202 0.58890 0.18079 END ------- Q.309 - Is AMORE giving the right answer? S.309 - Amore A.309 - In response to: I am trying out AMORE on a known system (solved structure R=19%) it seems to go fine single peaks high rms etc. Yet the solution appears wrong. My spacegroup is P41212 (#92) this is being correctly picked up by the program TRAFUN/ROTFUN. I apply the output solution to my centred (of mass) molecule. However this is related to the starting molecule by the following operation: ### CCP PROGRAM SUITE: LSQKAB VERSION 2.10: 03/06/94### WORKCD solved_refined_strucure.pdb REFRCD amore_sln.pdb TRANSLATION VECTOR IN fractions of cell edge 34.482246 34.511993 46.297478 (fraction 1/2, 1/2, 1/4) ROTATION MATRIX 1.000 0.000 0.000 0.000 -1.000 0.000 0.000 0.000 -1.000 This looks to me like x+1/2, -y+1/2, -z+1/4, which is not a symmetry operator for my spacegroup (whereas x+1/2, -y+1/2, -z+3/4 and -x+1/2, y+1/2, -z+1/4 are). However, x+1/2, -y+1/2, -z+1/4 is an operator for P43212. Does anyone have anys suggestions/comments? Answer, you have indeed picked up the solution for the symmetry related molecule. Your solution is 1 0 0 with translations (0.5,0.5,0.25) 0 -1 0 0 0 -1 You should notice that in P41212 there is an ambiguity in fixing the origin at (0,0,0), (0,0,0.5), (0.5,0.5,0) and (0.5,0.5,0.5). Applying an origin shift of (0,0,0.5) to your translation of (0.5,0.5,0.25) gives a solution which is related to the known structure by (x+1/2, -y+1/2, -z+3/4). A list of allowed origin translations for each space group are given in International Tables, Volume B, pp203-206. ------- (V) REFINEMENT Q.401 - How do I keep my Free R set intact? S.401 - Free_R A.401 - In repsonse to: I have a question to Axel Brünger's Free R-value concept: It is easy to use Rfree in X-PLOR and in the CCP4 environment, but how is it possible to go from X-PLOR to least squares refinement in CCP4 without changing the randomly selected test data set? Such a change makes no sense in the Rfree concept. Answer, f2mtz (and mtz2various) know about the CCP4/X-PLOR free R conventions and so allow you to change formats whilst maintaining the free R indicators. See the note about CTYPOUT `X' in f2mtz doc. You can also use SFTOOLS, available from: ------- (VI) PHASE_IMPROVEMENT Q.501 - What programs are available for symmetry averaging? S.501 - Averaging_programs A.501 - DM (which is part of CCP4) and RAVE (from Uppsala) are both capable of performing (non-)crystallographic symmetry averaging and are compatible with the other programs in the Suite. DM (see Q.503) also has facilities to apply solvent flattening, protein density histogram matching, Sayre's equations and skeletonisation - all of which may significantly improve your electron density maps. ------- Q.502 - How do I generate a mask from a model (PDB) file? S.502 - Mask_generation A.502 - NCSMASK can make a mask from a PDB file. It also allows manipulation of the mask (smoothing, expanding, contracting, checking overlaps etc.). Allied to this is MAPMASK which allows maps to be converted to masks and manipulated. Another useful program is BONES2PDB. This program will convert the output from O/bones into a pdb file, which can then be used for mask generation in NCSMASK. ------- Q.503 - What programs are available for density modification? S.503 - Density_modification A.503 - DM is the current CCP4 density modification program. In which the user can apply - Solvent flattening (to flatten the solvent region - may use automatic or user supplied protein mask) - Histogram matching (to fit the distribution of density points in the protein to the standard distribution for proteins) - Sayre's equations (for extracting the phase relationship information from high resolution data) - Skeletonisation (to impose connectivity constraints on the density as an aid to phase improvement) ------- Q.504 - How do I use a MAMA mask in CCP4? S.504 - Mama_mask A.504 - The `mama2ccp4' program will do the conversion. (Old versions of the Uppsala `mapman' program may get it wrong.) ------- Q.505 - How do I generate a FOM for DM, from a MR solution? S.505 - MR_and_DM A.505 - To generate a FOM for DM, you need to run SIGMAA with the 'PARTIAL' keyword. You need to input Fobs with FC and PHIC and WCMB is the appropriate weight to use. Here is an example below; sigmaa hklin gmto_sfall.mtz hklout gmto_sigmaa.mtz << eof partial labin FP=FP SIGFP=SIGFP FC=FCmolr PHIC=PHICmolr eof dm hklin gmto_sigmaa.mtz hklout gmtodm.mtz << my-data SOLC 0.35 MODE SOLV HIST NCYCLE 10 LABIN FP=FP SIGFP=SIGFP PHIO=PHICmolr FOMO=WCMB my-data However, Eleanor Dodson points out that unless you edit the mask it will only cover the original MR molecule. The protein density that is outside the mask will tend to be squashed along with the density. This is also a problem when coming from MIR phases but at least you still have the MIR map in which the protein density has not been squashed. She suggests running REFMAC for a few cycles and then using the generated Fourier coefficients to calculate the map. Hopefully, from this you will be able to extend/alter the model. ------- (VII) ANALYSIS Q.601 - Where shall we put our Hydrogen atoms? S.601 - Hydrogen_positions A.601 - In response to: We would like to do some surface/cavity calculations using VOIDOO. But we would like to do them while considering Hydrogen atoms on the surface (to add to the vdw radii). We can use XPLOR to add hydrogen atoms but then other programs seem to have a fit with them and it is a pain in the butt to properly set up the scripts to do a simple thing. So, is there a utility in O or CCP4 (or anywhere else for that matter) that will ADD hydrogens and name them H to an existing .PDB file? We could not find one (MOLEMAN will only export hydrogen added in but will actually place them). Answer, hydrogen bonds data and hydrogen atom positions( with hbplus -o option) can be got with the program "HBPLUS" this is part of the procheck group of software, contact: Ian McDonald (PG) Biological Structure and Modelling Unit Department of Biochemistry and Molecular Biology University College London Gower Street LONDON WC1E 6BT UK Email mcdonald@uk.ac.ucl.bsm ------- Q.602 - After phase improvement using DM what sort of map do I plot? S.602 - DM_map A.602 - After using density modification you will have a modified phase and weight. These will be extra columns in your MTZ file and can be assigned by using LABOUT. To produce the map you need to use FFT, using the columns below; the native structure factor with sigma value, the phase and weight from DM e.g. LABIN F1=Fnative SIG1=SIGnative PHI=PHIDM W=FOMDM In general it is not applicable to generate 2Fo-Fc maps using Fc calculated by the dm procedure. Although this is of course dependent on the type of proceedure used. It is better to generate a weighted Fourier, the change in weight is just as important as the change in phase. ------- (VIII) PRESENTATION Q.701 - Is NPO the same as PLUTO? S.701 - Plotting_program A.701 - NPO is the replacement for PLUTO, including facilities of both PLUTO and ORTEP (a small molecule plotting program). Keyword input for NPO should be similar to PLUTO, and is detailed in the doc/man pages. ------- Q.702 - How do I analyse my water molecules? S.702 - Analysis A.702 - In response to: Is there any other CCP program than CONTACT to analyze the water molecules concerning their numbers of contacts to protein atoms? Answer, you can use the ACT or DISTANG programs. ------- Q.703 - My NPO contours are all black... S.703 - NPO_input A.703 - In response to: I can obtain black contours, with negative ones dashed, by using the lines below in NPO. CONTRS NEG 5 to 40 by 5 MODE BLACK BELOW 0 BLACK DASHED 4 2 0 How can I get coloured contours without using PLUTO? Answer, the following command file draws contours red dashed below zero and black above zero npo mapin xy_patt.xmap \ plot m.plt\ << eof TITL skewing test CELL 105.831 105.831 153.649 90.00000 90.00000 120.00000 MAP CONTRS NEG 150 to 2500 by 150 # you always needed the second line for the two modes MODE BLACK MODE BELOW 0 RED DASHED 4 2 0 SECTNS 30 30 1 PLOT Y eof # To see the colours on the screen with xplot84driver or to get colour PS files, you must also make sure that you are using the colour AppDefaults. These are contained in ccp4/x-windows/XCCPJIFFY/XCCPJiffy.AppDefaults. You must move the the correct bits of this file to a suitable directory, see ccp4/include/ccp4.setup-dist. You can also enter these parameters into the x-database by using 'xrdb -merge XCCPJiffy.AppDefaults'. You can then check these using 'xrdb -query'. See also the man pages for xplot84driver. ------- Q.704 - How to convert a PLOT84 file to Postscript for printing? S.704 - PLOT84_conversion A.704 - There is a program called PLTDEV which converts PLOT84 file into Tektronix, HPGL, or Postscript files. An alternative (for use on X-windows terminals) is xplot84driver which is a menu driven/interactive program which allows you to view the file and store it as Postscript for hardcopy. ------- Q.705 - How do I customise the jiffies XPLOT84DRIVER and XLOGGRAPH? S.705 - XCCPJiffy A.705 - Thanks to Jan Zelinka for the original of the following summary. XPLOT84DRIVER and XLOGGRAPH are two jiffies for interpreting data generated by CCP4 programs. o XPLOT84DRIVER (x84 - a Cambridge born nickname) interprets graphics coded in the form of plot84 format files. o XLOGGRAPH interprets text format marked tables buried in log files. Both applications belong to the same "application class" XCCPJiffy. How the graphics are interpreted depends on application resources set by the application resources file. A default application resources file is provided in the distribution as pub/ccp4/ccp4/x-windows/XCCPJIFFY/XCCPJiffy.AppDefaults and should be installed in a suitable directory (for example, as the file $CCP4_LIB/X11/app-defaults/XCCPJiffy where the filename is the name of the application class). Alternatively, "xrdb -merge _filename_" can be used to explicitly load the application resources file. If your programs experience problems with finding X-resources look for instance at: Nye.A, O'Reilly T.: X Toolkit Intrinsic Programming Manual, O'Reilley & Associates. Inc.(1990) ISBN 0-937175-62-5, pp. 300-302. In principle, XCCPJiffies work in two modes: o Pen mode (used in the xloggraph) - an entity (line etc.) is drawn by a pen with specified width and colour. o Color and width mode (used in the xplot84driver) - color and width are set separately as defined within the plot84 graphics metafile. In both cases number of pens/colours/widths is limited to 10. These modes are governed by different attributes specified in the application resources file. The XCCPJiffy application resources file shipped with the package is really only a raw material and should be processed by the C-preprocessor or manually to produce the B/W or colour display version. Subsequently, resources for the screen output should be edited: ! xloggraph specific CCPJiffy.graph*draw2d.background: white XCCPJiffy.graph*draw2d.pen1Color: black XCCPJiffy.graph*draw2d.pen1LineWidth: 1 .... XCCPJiffy.graph*draw2d.pen10Color: indigo XCCPJiffy.graph*draw2d.pen10LineWidth: 1 ! xplot84driver specific XCCPJiffy.plot84Picture*draw2d.background: white XCCPJiffy.plot84Picture*draw2d.color1: black .... XCCPJiffy.plot84Picture*draw2d.color10: indigo Similarly, changing resources for the postscript interpreter, different postcript interpretations can be achieved. Postscript settings are separated from screen settings mostly because of different colour - B/W requirements. The value of the attribute is the Postcript directive(s) called when the colour/line_width/pen change is required. ! Postscript "pen-mode" used in xloggraph XCCPJiffy*psPen1Attr: 1 setlinejoin 0.5 setlinewidth [] 0 setdash ..... XCCPJiffy*psPen10Attr: 1 setlinejoin 0.9 setlinewidth [4 12] 0 setdash ! Postscript "change line width & change color mode" used in xplot84driver XCCPJiffy*psBasicLineWidth: 0.5 XCCPJiffy*psColor1: 1 setlinejoin [] 0 setdash ..... XCCPJiffy*psColor10: 1 setlinejoin [6 10] 0 setdash Alternatively, colour postscript pictures for xloggraph: .... XCCPJiffy*psPen10Attr: 1 setlinejoin 1.5 setlinewidth [] 0 setdash gsave 1.0000 0.6470 0.0000 setrgbcolor and for xplot84driver: .... XCCPJiffy*psColor10: 1 setlinejoin 1.5 setlinewidth [] 0 setdash gsave 1.0000 0.6470 0.0000 setrgbcolor Resources directing the plot/print destination (set in installation) can be altered, e.g.: !!!! Resources set during install procedure !!!! XCCPJiffy*psPlotCommand: lp -s -dHP_LaserJet XCCPJiffy*textPrintCommand: lp -s -denscript Finally, the Tom Oldfield font description file path required for plot84driver should be set by, e.g.: XCCPJiffy*tomFontPath: /y/programs/xtal/ccp4/lib/nice_font.bin ------- (IX) MISCELLANEOUS (X) OPERATING SYSTEM PROBLEMS Q.901 What should I look out for when compiling under IRIX6.*. S.901 IRIX6 A.901 On SGs `f77` and `cc` have a dual mode of operation, under IRIX6.* because some machines are 32-bit machines and some are 64-bit machines. You must decide which type of machine you have and use the appropriate flags for `f77` and `cc`. If you have a 64-bit machine (e.g. R10000, R8000) then use `f77 -64 -mips4 ...` and `cc -64 -mips4 ...`. If you have a 32-bit machine (e.g. R2000, R3000) then use `f77 -32 [mips1|mips2] ...` and `cc -32 [-mips1|mips2] ...`. There are some machines that have a mixed architecture (e.g. R4000, R4400), in this case use `f77 -n32 -mips3 ...` and `cc -n32 -mips3 ...` If you are unsure then use -32 because 64-bit machines can still compile 32-bit code. Make sure that the C code is compatible with the FORTRAN code otherwise you will get compiler errors like ar: Warning:This archive ONLY holds 64 bit objects. library.o is a 32 bit object file so symbol table info would not be generated for it. These changes above should be made in the $CCP4/configure file in $XFFLAGS and $XCFLAGS. Remember that the syntax of other command line options change when using either -64 or -32. ------- ---END--- ********************************************************************** Paul D. Hempstead Krebs Institute Department of Molecular Biology University of Sheffield Sheffield UK E-mail: p.hempstead@sheffield.ac.uk **********************************************************************