Average maps (*not* NCS averaging)
Hi there, I'd like to average 2Fo-Fc maps from 2 (approximately) isomorphous data sets (one with ligand, one without) to see whether I get better density for some of the noisier bits of the structure. I don't want to do any kind of solvent flattening, histogram matching, cyclical averaging, etc; I just want to sum the equivalent pixels and divide by 2. Is it possible to do this easily using any of the phenix tools? It seems like all the required plumbing is there with get_cc_mtz_mtz or superpose_maps, but I can't find a phenix utility for averaging two superposed maps. Thanks, Stephen -- Dr Stephen Graham 1851 Research Fellow Cambridge Institute for Medical Research Wellcome Trust/MRC Building Addenbrooke's Hospital, Hills Road Cambridge, CB2 0XY, UK Phone: +44 1223 762 638
Stephen, I believe that the Uppsala program COMAP will do what you want, assuming you have maps in CCP4 format. It's part of the Uppsala averaging suite. Phil Jeffrey Princeton On 3/10/11 12:15 PM, Stephen Graham wrote:
Hi there,
I'd like to average 2Fo-Fc maps from 2 (approximately) isomorphous data sets (one with ligand, one without) to see whether I get better density for some of the noisier bits of the structure. I don't want to do any kind of solvent flattening, histogram matching, cyclical averaging, etc; I just want to sum the equivalent pixels and divide by 2.
Is it possible to do this easily using any of the phenix tools? It seems like all the required plumbing is there with get_cc_mtz_mtz or superpose_maps, but I can't find a phenix utility for averaging two superposed maps.
Thanks,
Stephen
Hi Stephen, I don't know of any phenix tools that will average two maps, though you are right that everything is there to do it quite easily. -Tom On Mar 10, 2011, at 10:15 AM, Stephen Graham wrote:
Hi there,
I'd like to average 2Fo-Fc maps from 2 (approximately) isomorphous data sets (one with ligand, one without) to see whether I get better density for some of the noisier bits of the structure. I don't want to do any kind of solvent flattening, histogram matching, cyclical averaging, etc; I just want to sum the equivalent pixels and divide by 2.
Is it possible to do this easily using any of the phenix tools? It seems like all the required plumbing is there with get_cc_mtz_mtz or superpose_maps, but I can't find a phenix utility for averaging two superposed maps.
Thanks,
Stephen
-- Dr Stephen Graham 1851 Research Fellow Cambridge Institute for Medical Research Wellcome Trust/MRC Building Addenbrooke's Hospital, Hills Road Cambridge, CB2 0XY, UK Phone: +44 1223 762 638 _______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
Thomas C. Terwilliger Mail Stop M888 Los Alamos National Laboratory Los Alamos, NM 87545 Tel: 505-667-0072 email: [email protected] Fax: 505-665-3024 SOLVE web site: http://solve.lanl.gov PHENIX web site: http:www.phenix-online.org ISFI Integrated Center for Structure and Function Innovation web site: http://techcenter.mbi.ucla.edu TB Structural Genomics Consortium web site: http://www.doe-mbi.ucla.edu/TB CBSS Center for Bio-Security Science web site: http://www.lanl.gov/cbss
Hi, look for " def combine_maps" here: http://cci.lbl.gov/~afonine/mssa2/run.py that does it what you want. It reads in all files with map coefficients, averages them, and outputs averaged map coefficients. Since Fourier transformation is a linear operation, averaging map coefficients is equivalent to averaging actual maps. In 5 minutes I can turn this script into something like phenix.average_map_coefficients map_coeff1 map_coeff2.mtz labels=" 2FOFCWT" Do we need this tool? Tom, how much of non-isomorphism we should allow before the above tool stops? Pavel. On 3/10/11 9:49 AM, Tom Terwilliger wrote:
Hi Stephen, I don't know of any phenix tools that will average two maps, though you are right that everything is there to do it quite easily. -Tom
On Mar 10, 2011, at 10:15 AM, Stephen Graham wrote:
Hi there,
I'd like to average 2Fo-Fc maps from 2 (approximately) isomorphous data sets (one with ligand, one without) to see whether I get better density for some of the noisier bits of the structure. I don't want to do any kind of solvent flattening, histogram matching, cyclical averaging, etc; I just want to sum the equivalent pixels and divide by 2.
Is it possible to do this easily using any of the phenix tools? It seems like all the required plumbing is there with get_cc_mtz_mtz or superpose_maps, but I can't find a phenix utility for averaging two superposed maps.
Thanks,
Stephen
-- Dr Stephen Graham 1851 Research Fellow Cambridge Institute for Medical Research Wellcome Trust/MRC Building Addenbrooke's Hospital, Hills Road Cambridge, CB2 0XY, UK Phone: +44 1223 762 638 _______________________________________________ phenixbb mailing list [email protected] mailto:[email protected] http://phenix-online.org/mailman/listinfo/phenixbb
Thomas C. Terwilliger Mail Stop M888 Los Alamos National Laboratory Los Alamos, NM 87545
Tel: 505-667-0072 email: [email protected] mailto:[email protected] Fax: 505-665-3024 SOLVE web site: http://solve.lanl.gov PHENIX web site: http:www.phenix-online.org ISFI Integrated Center for Structure and Function Innovation web site: http://techcenter.mbi.ucla.edu TB Structural Genomics Consortium web site: http://www.doe-mbi.ucla.edu/TB CBSS Center for Bio-Security Science web site: http://www.lanl.gov/cbss
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
On Thu, Mar 10, 2011 at 9:15 AM, Stephen Graham
I'd like to average 2Fo-Fc maps from 2 (approximately) isomorphous data sets (one with ligand, one without) to see whether I get better density for some of the noisier bits of the structure. I don't want to do any kind of solvent flattening, histogram matching, cyclical averaging, etc; I just want to sum the equivalent pixels and divide by 2.
Is it possible to do this easily using any of the phenix tools? It seems like all the required plumbing is there with get_cc_mtz_mtz or superpose_maps, but I can't find a phenix utility for averaging two superposed maps.
I suspect that this is possible in Resolve, but I don't know what the protocol is - at any rate, there isn't a simple utility for this. I have a script that does most of what you need, but I haven't had time to turn it into a useful multi-purpose command. Since it's useful for other applications (such as multi-start SA), I had planned to do this in the near future. But it sounds like COMAP may be your best bet for now. -Nat
On 17:15 Thu 10 Mar , Stephen Graham wrote:
Hi there,
I'd like to average 2Fo-Fc maps from 2 (approximately) isomorphous data sets (one with ligand, one without) to see whether I get better density for some of the noisier bits of the structure. I don't want to do any kind of solvent flattening, histogram matching, cyclical averaging, etc; I just want to sum the equivalent pixels and divide by 2.
Is it possible to do this easily using any of the phenix tools? It seems like all the required plumbing is there with get_cc_mtz_mtz or superpose_maps, but I can't find a phenix utility for averaging two superposed maps.
Well, Coot is distributed with phenix so I guess you could call it a phenix tool. You can use tools in Extensions->Maps to first superpose the maps (transform by LSQ fit), then average them. -- Thanks, Donnie Donald S. Berkholz, Ph.D. Research Fellow James R. Thompson lab, Physiology & Biomedical Engineering Grazia Isaya lab, Pediatric & Adolescent Medicine Medical Sciences 2-66 Mayo Clinic College of Medicine 200 First Street SW Rochester, MN 55905 office: 507-538-6924 cell: 612-991-1321
Donnie, Coot is not a part of PHENIX and is not distributed with PHENIX. PHENIX can use Coot if it is available (installed separately by the user). Pavel. On 3/10/11 7:26 PM, Donnie Berkholz wrote:
Hi there,
I'd like to average 2Fo-Fc maps from 2 (approximately) isomorphous data sets (one with ligand, one without) to see whether I get better density for some of the noisier bits of the structure. I don't want to do any kind of solvent flattening, histogram matching, cyclical averaging, etc; I just want to sum the equivalent pixels and divide by 2.
Is it possible to do this easily using any of the phenix tools? It seems like all the required plumbing is there with get_cc_mtz_mtz or superpose_maps, but I can't find a phenix utility for averaging two superposed maps. Well, Coot is distributed with phenix so I guess you could call it a
On 17:15 Thu 10 Mar , Stephen Graham wrote: phenix tool. You can use tools in Extensions->Maps to first superpose the maps (transform by LSQ fit), then average them.
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
On 20:41 Thu 10 Mar , Pavel Afonine wrote:
Donnie,
Coot is not a part of PHENIX and is not distributed with PHENIX. PHENIX can use Coot if it is available (installed separately by the user).
Thanks for the clarification. I guess if I had said "integrated with phenix," it would have been right. The important part of the answer remains.. -- Thanks, Donnie Donald S. Berkholz, Ph.D. Research Fellow James R. Thompson lab, Physiology & Biomedical Engineering Grazia Isaya lab, Pediatric & Adolescent Medicine Medical Sciences 2-66 Mayo Clinic College of Medicine 200 First Street SW Rochester, MN 55905 office: 507-538-6924 cell: 612-991-1321
Hi, I just want to point out that for averaging maps (i.e. averaging between crystals instead of averaging within a single crystal as is usually done), the 2 maps to be averaged do not necessarily have the same distribution of density, the same density histogram, the same dynamic range if you prefer. Averaging 2 maps which do not have the same histograms without "real space scaling" by histogram matching (or mapping) can lead to map deterioration instead of map improvement. See the example given in Methods in Enzymology 277, pp18-53, the figure showing this is on pages 36-37. Fred. Stephen Graham wrote:
Hi there,
I'd like to average 2Fo-Fc maps from 2 (approximately) isomorphous data sets (one with ligand, one without) to see whether I get better density for some of the noisier bits of the structure. I don't want to do any kind of solvent flattening, histogram matching, cyclical averaging, etc; I just want to sum the equivalent pixels and divide by 2.
Is it possible to do this easily using any of the phenix tools? It seems like all the required plumbing is there with get_cc_mtz_mtz or superpose_maps, but I can't find a phenix utility for averaging two superposed maps.
Thanks,
Stephen
I'm not sure this is a particularly rigorous way to get "noisier bits" less noisy. Surely the way to do it is to refine one model using the other model as restraints. Both refmac and buster do it, (maybe phenix.refine too). Cheers phx On 10/03/2011 17:15, Stephen Graham wrote:
Hi there,
I'd like to average 2Fo-Fc maps from 2 (approximately) isomorphous data sets (one with ligand, one without) to see whether I get better density for some of the noisier bits of the structure. I don't want to do any kind of solvent flattening, histogram matching, cyclical averaging, etc; I just want to sum the equivalent pixels and divide by 2.
Is it possible to do this easily using any of the phenix tools? It seems like all the required plumbing is there with get_cc_mtz_mtz or superpose_maps, but I can't find a phenix utility for averaging two superposed maps.
Thanks,
Stephen
Dear all,
Thanks for the many responses.
I can easily generate superposed maps using phenix.superpose_maps.
Unfortunately this doesn't create maps with the same origin and
extents, etc., thus neither COMAP nor any of the CCP4 tools I've found
will average them trivially (I was trying to be lazy and get around
having to get the maps onto the same sampling grid...).
Donnie: Can COOT average arbitrary maps? I could not find that in the
online documentation (just NCS averaging).
Frank: Agreed when you have a high res 'good' structure and a low res
'bad' one. I have two high res data sets where a loop is modelled
(incorrectly) in both and am trying to clean up the density for
building, rather than restrain a wayward structure to a known ideal
model.
Pavel: Unfortunately your script dies with array bounds errors:
else: data += ma.data()
RuntimeError: scitbx array_family range error
Is this because it assumes the two mtz files have the same resolution
(i.e. len(ma) is the same for all)?
I was not fully aware of the caveats Fred mentioned - if histogram
matching is a must I'm probably better of doing this 'properly' using
resolve/dmmulti/etc.
Again, thanks for all the comments,
Stephen
On 11 March 2011 08:14, Frank von Delft
I'm not sure this is a particularly rigorous way to get "noisier bits" less noisy. Surely the way to do it is to refine one model using the other model as restraints. Both refmac and buster do it, (maybe phenix.refine too).
Cheers phx
On 10/03/2011 17:15, Stephen Graham wrote:
Hi there,
I'd like to average 2Fo-Fc maps from 2 (approximately) isomorphous data sets (one with ligand, one without) to see whether I get better density for some of the noisier bits of the structure. I don't want to do any kind of solvent flattening, histogram matching, cyclical averaging, etc; I just want to sum the equivalent pixels and divide by 2.
Is it possible to do this easily using any of the phenix tools? It seems like all the required plumbing is there with get_cc_mtz_mtz or superpose_maps, but I can't find a phenix utility for averaging two superposed maps.
Thanks,
Stephen
-- Dr Stephen Graham 1851 Research Fellow Cambridge Institute for Medical Research Wellcome Trust/MRC Building Addenbrooke's Hospital, Hills Road Cambridge, CB2 0XY, UK Phone: +44 1223 762 638
Hello Stephen-
I may be wrong but it seems like you are just doing averaging between two crystals. If so can you just use MAVE from the Uppsala software package RAVE? I'm pretty sure i did this with help from MAPMAN(to help with the issues of extent, etc). I'll look for a script when I get to work.
- Todd
-----Original Message-----
From: [email protected] on behalf of Stephen Graham
Sent: Fri 3/11/2011 5:29 AM
To: PHENIX user mailing list
Subject: Re: [phenixbb] Average maps (*not* NCS averaging)
Dear all,
Thanks for the many responses.
I can easily generate superposed maps using phenix.superpose_maps.
Unfortunately this doesn't create maps with the same origin and
extents, etc., thus neither COMAP nor any of the CCP4 tools I've found
will average them trivially (I was trying to be lazy and get around
having to get the maps onto the same sampling grid...).
Donnie: Can COOT average arbitrary maps? I could not find that in the
online documentation (just NCS averaging).
Frank: Agreed when you have a high res 'good' structure and a low res
'bad' one. I have two high res data sets where a loop is modelled
(incorrectly) in both and am trying to clean up the density for
building, rather than restrain a wayward structure to a known ideal
model.
Pavel: Unfortunately your script dies with array bounds errors:
else: data += ma.data()
RuntimeError: scitbx array_family range error
Is this because it assumes the two mtz files have the same resolution
(i.e. len(ma) is the same for all)?
I was not fully aware of the caveats Fred mentioned - if histogram
matching is a must I'm probably better of doing this 'properly' using
resolve/dmmulti/etc.
Again, thanks for all the comments,
Stephen
On 11 March 2011 08:14, Frank von Delft
I'm not sure this is a particularly rigorous way to get "noisier bits" less noisy. Surely the way to do it is to refine one model using the other model as restraints. Both refmac and buster do it, (maybe phenix.refine too).
Cheers phx
On 10/03/2011 17:15, Stephen Graham wrote:
Hi there,
I'd like to average 2Fo-Fc maps from 2 (approximately) isomorphous data sets (one with ligand, one without) to see whether I get better density for some of the noisier bits of the structure. I don't want to do any kind of solvent flattening, histogram matching, cyclical averaging, etc; I just want to sum the equivalent pixels and divide by 2.
Is it possible to do this easily using any of the phenix tools? It seems like all the required plumbing is there with get_cc_mtz_mtz or superpose_maps, but I can't find a phenix utility for averaging two superposed maps.
Thanks,
Stephen
-- Dr Stephen Graham 1851 Research Fellow Cambridge Institute for Medical Research Wellcome Trust/MRC Building Addenbrooke's Hospital, Hills Road Cambridge, CB2 0XY, UK Phone: +44 1223 762 638 _______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb This email was scanned with Mcafee's Anti-Virus appliance, but this is no guarantee that no virus exists. You are asked to make sure you have virus protection and that it is up to date.
On 11:29 Fri 11 Mar , Stephen Graham wrote:
Donnie: Can COOT average arbitrary maps? I could not find that in the online documentation (just NCS averaging).
Yes, I even pointed you to the relevant menu entries. I've previously used this to average maps from crystals in two different space groups. I don't think it does anything smart to give you a complete unit cell, though — just a chunk of map near your protein — but I could be wrong. -- Thanks, Donnie Donald S. Berkholz, Ph.D. Research Fellow James R. Thompson lab, Physiology & Biomedical Engineering Grazia Isaya lab, Pediatric & Adolescent Medicine Medical Sciences 2-66 Mayo Clinic College of Medicine 200 First Street SW Rochester, MN 55905 office: 507-538-6924 cell: 612-991-1321
Yes, I even pointed you to the relevant menu entries. I've previously used this to average maps from crystals in two different space groups.
Ah - so there is! My bad. This is the perfect quick-and-dirty solution I was looking for...it even gives some (subjectively) slightly more interpretable density. Now off to tackle this bad loop... Thanks, Stephen -- Dr Stephen Graham 1851 Research Fellow Cambridge Institute for Medical Research Wellcome Trust/MRC Building Addenbrooke's Hospital, Hills Road Cambridge, CB2 0XY, UK Phone: +44 1223 762 638
participants (9)
-
Donnie Berkholz -
Frank von Delft -
Green, Todd -
Nathaniel Echols -
Pavel Afonine -
Phil Jeffrey -
Stephen Graham -
Tom Terwilliger -
Vellieux Frederic