Re: [COOT] Automatic column asignment for refmac
Hi! Coot has to be told what columns to use for different programs. We've coded fore refmac, phenix.refine and my density modification programs, but not for phaser because no-one has asked. Can you send me a dump of the mtz header with a list of the columns? Which columns would you like used by default on auto open? Should there be a difference map as well? Thanks, Kevin alexander.schif...@sanofi-aventis.com wrote: Dear Paul, There seems to be a problem with the automatic asignment for F and SIGF when using a file from phaser (processed by autoproc). The FCTR column seems to be used instead of F but there is no reference what has been used in the logfile (sample attached). Unfortunately the only indication something went wrong is the high R-Factor, which in this case could also be due to the fact that the MR solution is incorrect. Best rega <> rds, Alexander Mit freundlichen Grüßen / Best regards / Cordialement Dr. Alexander Schiffer Sanofi-Aventis Deutschland GmbH R&D CAS Structural Biology FFM Industriepark Hoechst Bldg. G877, Room 029 D-65926 Frankfurt am Main t: +49 69 305 24896 f: +49 69 305 80169 w:www.sanofi-aventis.de * Sanofi-Aventis Deutschland GmbH · Sitz der Gesellschaft: Frankfurt am Main · Handelsregister: Frankfurt am Main, Abt. B Nr. 40661 Vorsitzender des Aufsichtsrats: Hanspeter Spek - Geschäftsführer: Dr. Martin Siewert (Vorsitzender), Ulf Bialojahn, Dr. Matthias Braun, Peter Guenter, Prof. Dr. Dr. Werner Kramer, Dr. Klaus Menken, Dr. Heinz Riederer *
Re: [COOT] Automatic column asignment for refmac
Dear Alexander, FCTR looks like it is coming from a BUSTER refinement: http://www.globalphasing.com/buster/manual/buster/manual/chapter5.html#mlphas This column is obviously only intended for looking at maps (however: usually one looks at the 2FOFCWT/PH2FOFCWT and FOFCWT/PHFOFCWT maps). So there mus thave been some additional step like autoPROC -> Phaser -> BUSTER -> Coot or autoPROC -> BUSTER -> Phaser -> Coot involved? Those problems often occur when one keeps 'chaining' the use of MTZ files: the MTZ file for looking at maps (all kind of additional columns, scaling might have been taken place etc) should ideally be different from the one used in refinement (this should always be the same 'master' file after data processing and adding test set flag). I guess by combining map viewing in Coot with driving refinement from the gui as well this can become tricky. Cheers Clemens On Mon, Nov 30, 2009 at 10:23:26AM +0100, alexander.schif...@sanofi-aventis.com wrote: > Dear Paul, > > There seems to be a problem with the automatic asignment for F and SIGF when > using a file from phaser (processed by autoproc). The FCTR column seems to be > used instead of F but there is no reference what has been used in the logfile > (sample attached). Unfortunately the only indication something went wrong is > the high R-Factor, which in this case could also be due to the fact that the > MR solution is incorrect. > > Best rega <> rds, > > Alexander > > > Mit freundlichen Grüßen / Best regards / Cordialement > > Dr. Alexander Schiffer > > Sanofi-Aventis Deutschland GmbH > R&D CAS Structural Biology FFM > Industriepark Hoechst > Bldg. G877, Room 029 > D-65926 Frankfurt am Main > t: +49 69 305 24896 > f: +49 69 305 80169 > w:www.sanofi-aventis.de > > * > Sanofi-Aventis Deutschland GmbH · Sitz der Gesellschaft: Frankfurt am Main · > Handelsregister: Frankfurt am Main, Abt. B Nr. 40661 > Vorsitzender des Aufsichtsrats: Hanspeter Spek - Geschäftsführer: Dr. Martin > Siewert (Vorsitzender), Ulf Bialojahn, Dr. Matthias Braun, > Peter Guenter, Prof. Dr. Dr. Werner Kramer, Dr. Klaus Menken, Dr. Heinz > Riederer > * > Content-Description: refmac-from-coot-0.log > sh: octet-filter: command not found
Re: [COOT] Automatic column asignment for refmac
Hi Kevin, As Clemens pointed out it was not a phaser but a user problem. I accidently used a buster mtz file for phaser. For autoproc - phaser everything works as expected. I have attached the mtzdump of a buster mtz-file. If you could support auto-open for that it would be great. Do you see a way to include in the refmac-from-coot logfile which mtz columns were used? Alexander -Original Message- From: Kevin Cowtan [mailto:cow...@ysbl.york.ac.uk] Sent: Monday, November 30, 2009 11:02 AM To: Schiffer, Alexander R&D/DE Cc: COOT@JISCMAIL.AC.UK Subject: Re: [COOT] Automatic column asignment for refmac Hi! Coot has to be told what columns to use for different programs. We've coded fore refmac, phenix.refine and my density modification programs, but not for phaser because no-one has asked. Can you send me a dump of the mtz header with a list of the columns? Which columns would you like used by default on auto open? Should there be a difference map as well? Thanks, Kevin alexander.schif...@sanofi-aventis.com wrote: > Dear Paul, > > There seems to be a problem with the automatic asignment for F and SIGF when > using a file from phaser (processed by autoproc). The FCTR column seems to be > used instead of F but there is no reference what has been used in the logfile > (sample attached). Unfortunately the only indication something went wrong is > the high R-Factor, which in this case could also be due to the fact that the > MR solution is incorrect. > > Best rega <> rds, > > Alexander > > > Mit freundlichen Grüßen / Best regards / Cordialement > > Dr. Alexander Schiffer > > Sanofi-Aventis Deutschland GmbH > R&D CAS Structural Biology FFM > Industriepark Hoechst > Bldg. G877, Room 029 > D-65926 Frankfurt am Main > t: +49 69 305 24896 > f: +49 69 305 80169 > w:www.sanofi-aventis.de > > ** > ** > * Sanofi-Aventis Deutschland GmbH · Sitz der > Gesellschaft: Frankfurt am Main · Handelsregister: Frankfurt am Main, > Abt. B Nr. 40661 Vorsitzender des Aufsichtsrats: Hanspeter Spek - > Geschäftsführer: Dr. Martin Siewert (Vorsitzender), Ulf Bialojahn, Dr. > Matthias Braun, Peter Guenter, Prof. Dr. Dr. Werner Kramer, Dr. Klaus > Menken, Dr. Heinz Riederer > ** > ** > * > refine.mtzlog Description: refine.mtzlog
Re: [COOT] Automatic column asignment for refmac
Hi, the automatic selection of structure factors for refinement happens in Refmac and not in Coot. This is one for Garib. I dont know exactly how Refmac chooses which structure factors to use, but it certainly should write the selected ones in the log file (maybe newer versions of refmac already do so?!). If you want to make sure Refmac uses the structure factors you want, then select the corresponding option and do not trust Refmac to do the job for you. Furthermore you probably should work with the original mtz file when doing refinement. In that case there wont be too many different options to pick the 'wrong' structure factors for refinement as well. Hope this helps, B P.S. Kevin, I guess you misunderstood the problem. It's about SF used in refmac (via Coot) rather than (auto)open mtz files. Dear Paul, There seems to be a problem with the automatic asignment for F and SIGF when using a file from phaser (processed by autoproc). The FCTR column seems to be used instead of F but there is no reference what has been used in the logfile (sample attached). Unfortunately the only indication something went wrong is the high R-Factor, which in this case could also be due to the fact that the MR solution is incorrect. Best rega <> rds, Alexander Mit freundlichen Grüßen / Best regards / Cordialement Dr. Alexander Schiffer Sanofi-Aventis Deutschland GmbH R&D CAS Structural Biology FFM Industriepark Hoechst Bldg. G877, Room 029 D-65926 Frankfurt am Main t: +49 69 305 24896 f: +49 69 305 80169 w:www.sanofi-aventis.de * Sanofi-Aventis Deutschland GmbH · Sitz der Gesellschaft: Frankfurt am Main · Handelsregister: Frankfurt am Main, Abt. B Nr. 40661 Vorsitzender des Aufsichtsrats: Hanspeter Spek - Geschäftsführer: Dr. Martin Siewert (Vorsitzender), Ulf Bialojahn, Dr. Matthias Braun, Peter Guenter, Prof. Dr. Dr. Werner Kramer, Dr. Klaus Menken, Dr. Heinz Riederer * -- *** Dr. Bernhard Lohkamp Assistant Professor Div. Molecular Structural Biology Dept. of Medical Biochemistry and Biophysics (MBB) Karolinska Institutet S-17177 Stockholm Sweden phone: (+46) 08-52487673 fax: (+46) 08-327626 email: bernhard.lohk...@ki.se
Re: [COOT] Automatic column asignment for refmac
OK, so the map coefficients are: 1. FCTR PHICTR 2. 2FOFCWT PH2FOFCWT 3. FOFCWT PHFOFCWT 4. FOFRGSLV PHFOFRGSLV 5. FC PHIC 6. FOSC SIGFOSC I think that auto-open (assuming you are using a 0.6pre from the last 6 months) should automatically pick up (2) and (3) for a normal and difference map - is that what you see? Which other columns would you like maps from? Just (1)? Do you see a way to include in the refmac-from-coot logfile which mtz columns were used? Sorry, didn't understand this bit. K
Re: [COOT] Automatic column asignment for refmac
Bernhard is right, sorry. Ignore me! Kevin Bernhard Lohkamp wrote: Hi, the automatic selection of structure factors for refinement happens in Refmac and not in Coot. This is one for Garib. I dont know exactly how Refmac chooses which structure factors to use, but it certainly should write the selected ones in the log file (maybe newer versions of refmac already do so?!). If you want to make sure Refmac uses the structure factors you want, then select the corresponding option and do not trust Refmac to do the job for you. Furthermore you probably should work with the original mtz file when doing refinement. In that case there wont be too many different options to pick the 'wrong' structure factors for refinement as well. Hope this helps, B P.S. Kevin, I guess you misunderstood the problem. It's about SF used in refmac (via Coot) rather than (auto)open mtz files. Dear Paul, There seems to be a problem with the automatic asignment for F and SIGF when using a file from phaser (processed by autoproc). The FCTR column seems to be used instead of F but there is no reference what has been used in the logfile (sample attached). Unfortunately the only indication something went wrong is the high R-Factor, which in this case could also be due to the fact that the MR solution is incorrect. Best rega <> rds, Alexander Mit freundlichen Grüßen / Best regards / Cordialement Dr. Alexander Schiffer Sanofi-Aventis Deutschland GmbH R&D CAS Structural Biology FFM Industriepark Hoechst Bldg. G877, Room 029 D-65926 Frankfurt am Main t: +49 69 305 24896 f: +49 69 305 80169 w:www.sanofi-aventis.de * Sanofi-Aventis Deutschland GmbH · Sitz der Gesellschaft: Frankfurt am Main · Handelsregister: Frankfurt am Main, Abt. B Nr. 40661 Vorsitzender des Aufsichtsrats: Hanspeter Spek - Geschäftsführer: Dr. Martin Siewert (Vorsitzender), Ulf Bialojahn, Dr. Matthias Braun, Peter Guenter, Prof. Dr. Dr. Werner Kramer, Dr. Klaus Menken, Dr. Heinz Riederer *
Re: [COOT] Automatic column asignment for refmac
At the moment refmac does extremely simple selection. First structure factor label and the first sigma that has similar name to the first structure factor, if not then the first sigma. It works for a class of cases but it cannot be universal. While we are on this matter I think it would be good to not mix input mtz and output mtz from refinement. They may have completely different meaning. I.e. for refinement always original mtz after scala/truncate with correct space group assigned should be used. Structure factors in the output mtz may have been modified by the programs and these modifications are based on some assumptions that may or may not be correct at the time of that program processing that dataset. As information changes (e.g. better model) modification may also change. i hope I did not add more confusion. Garib On 30 Nov 2009, at 10:43, Bernhard Lohkamp wrote: Hi, the automatic selection of structure factors for refinement happens in Refmac and not in Coot. This is one for Garib. I dont know exactly how Refmac chooses which structure factors to use, but it certainly should write the selected ones in the log file (maybe newer versions of refmac already do so?!). If you want to make sure Refmac uses the structure factors you want, then select the corresponding option and do not trust Refmac to do the job for you. Furthermore you probably should work with the original mtz file when doing refinement. In that case there wont be too many different options to pick the 'wrong' structure factors for refinement as well. Hope this helps, B P.S. Kevin, I guess you misunderstood the problem. It's about SF used in refmac (via Coot) rather than (auto)open mtz files. Dear Paul, There seems to be a problem with the automatic asignment for F and SIGF when using a file from phaser (processed by autoproc). The FCTR column seems to be used instead of F but there is no reference what has been used in the logfile (sample attached). Unfortunately the only indication something went wrong is the high R-Factor, which in this case could also be due to the fact that the MR solution is incorrect. Best rega <> rds, Alexander Mit freundlichen Grüßen / Best regards / Cordialement Dr. Alexander Schiffer Sanofi-Aventis Deutschland GmbH R&D CAS Structural Biology FFM Industriepark Hoechst Bldg. G877, Room 029 D-65926 Frankfurt am Main t: +49 69 305 24896 f: +49 69 305 80169 w:www.sanofi-aventis.de * Sanofi-Aventis Deutschland GmbH · Sitz der Gesellschaft: Frankfurt am Main · Handelsregister: Frankfurt am Main, Abt. B Nr. 40661 Vorsitzender des Aufsichtsrats: Hanspeter Spek - Geschäftsführer: Dr. Martin Siewert (Vorsitzender), Ulf Bialojahn, Dr. Matthias Braun, Peter Guenter, Prof. Dr. Dr. Werner Kramer, Dr. Klaus Menken, Dr. Heinz Riederer * -- *** Dr. Bernhard Lohkamp Assistant Professor Div. Molecular Structural Biology Dept. of Medical Biochemistry and Biophysics (MBB) Karolinska Institutet S-17177 Stockholm Sweden phone: (+46) 08-52487673 fax: (+46) 08-327626 email: bernhard.lohk...@ki.se
Re: [COOT] Automatic column asignment for refmac
Alexander, Do you see a way to include in the refmac-from-coot logfile which mtz columns were used? No, since, as mentioned earlier, this is a refmac decision. However I agree that the refmac log file should tell you which column was used when automatically picking SF (maybe it does already, Garib?). B -- *** Dr. Bernhard Lohkamp Assistant Professor Div. Molecular Structural Biology Dept. of Medical Biochemistry and Biophysics (MBB) Karolinska Institutet S-17177 Stockholm Sweden phone: (+46) 08-52487673 fax: (+46) 08-327626 email: bernhard.lohk...@ki.se
Re: [COOT] Automatic column asignment for refmac
Yes. It should and I will add it now. I do not think refmac writes this info into the output file but it will do in one or two days time. GArib On 30 Nov 2009, at 11:32, Bernhard Lohkamp wrote: Alexander, Do you see a way to include in the refmac-from-coot logfile which mtz columns were used? No, since, as mentioned earlier, this is a refmac decision. However I agree that the refmac log file should tell you which column was used when automatically picking SF (maybe it does already, Garib?). B -- *** Dr. Bernhard Lohkamp Assistant Professor Div. Molecular Structural Biology Dept. of Medical Biochemistry and Biophysics (MBB) Karolinska Institutet S-17177 Stockholm Sweden phone: (+46) 08-52487673 fax: (+46) 08-327626 email: bernhard.lohk...@ki.se
Re: [COOT] Automatic column asignment for refmac
Hi Kevin, just for the record (even if the original question/thread was about something slightly different): On Mon, Nov 30, 2009 at 11:13:14AM +, Kevin Cowtan wrote: > OK, so the map coefficients are: > > 1. FCTR PHICTR > 2. 2FOFCWT PH2FOFCWT > 3. FOFCWT PHFOFCWT > 4. FOFRGSLV PHFOFRGSLV > 5. FC PHIC > 6. FOSC SIGFOSC Those are not map coefficients (these are scaled Fobs/SIGFobs). > I think that auto-open (assuming you are using a 0.6pre from the last 6 > months) should automatically pick up (2) and (3) for a normal and > difference map - is that what you see? This has always worked as expected, yes. Running % coot --pdb refine.pdb --auto refine.mtz after BUSTER works perfectly for me every time ;-) Cheers Clemens -- *** * Clemens Vonrhein, Ph.D. vonrhein AT GlobalPhasing DOT com * * Global Phasing Ltd. * Sheraton House, Castle Park * Cambridge CB3 0AX, UK *-- * BUSTER Development Group (http://www.globalphasing.com) ***