Refinement using data with pseudo translational symmetry
To all, First I apologize for the long email, but I wont to make sure that I give enough information to describe the problem. I have been working on a structure of two proteins in complex with each other (one forms a homodimer, with each monomer bound to 1 monomer of the other protein), using data, that according to phenix.xtriage contains pseudo translational symmetry (output pasted below). I have done a lot of searching of both the phenixBB and ccp4BB regarding solutions to this problem, but unfortunately most responses seem to be directed at resolving issues with MR. I have successfully performed MR using both phenix phaser, ccp4 phaser (the most updated version), as well as molrep. The issue arises upon structure refinement, where my Rwork and Rfree are essentially stuck at 28 and 33, respectively. I have built the entire structure, including ligands, except for any solvent molecules. Here are the details for the data/structure: SG: P212121 Cell: 72 124 175 90 90 90 Res: 50-2.8 Patterson analyses ------------------ Largest Patterson peak with length larger than 15 Angstrom Frac. coord. : 0.155 0.000 0.500 Distance to origin : 87.222 Height (origin=100) : 34.517 p_value(height) : 6.739e-04 The reported p_value has the following meaning: The probability that a peak of the specified height or larger is found in a Patterson function of a macro molecule that does not have any translational pseudo symmetry is equal to 6.739e-04. p_values smaller than 0.05 might indicate weak translational pseudo symmetry, or the self vector of a large anomalous scatterer such as Hg, whereas values smaller than 1e-3 are a very strong indication for the presence of translational pseudo symmetry. Xtriage notes that the if the PTS is crystallographic, that C2221 is a possible SG (x+1/6, y, z+1/2). However, neither XDS or HKL picks orthorhombic C, and the indexing/integrating doesn't work if I force it. I should also not there is no twinning detected by either xtriage or twinning servers. I have done a number of things to try to resolve this: -rescaling in lower symmetry (P21 and P1) -rescaling in the PG P222 and letting the MR program decide the proper space group -shifting origin and re-refining -a number of different refinement protocols (TLS, optimized weights/adp, simulated annealing, several cycles of rigid body, etc.) I have used HKL2000 and XDS to index/integrate/scale the data, both yield the same results, with slightly different completeness and Rmerge values, but refining with either gives similar R factor values. Does any know of any other possible things I should try to refine this data? I am happy to provide additional information upon request. Thanks in advance for any help! Bret
On Sun, Mar 18, 2012 at 8:37 AM, Bret Wallace
I have done a number of things to try to resolve this:
-rescaling in lower symmetry (P21 and P1) -rescaling in the PG P222 and letting the MR program decide the proper space group -shifting origin and re-refining -a number of different refinement protocols (TLS, optimized weights/adp, simulated annealing, several cycles of rigid body, etc.)
What did you do about NCS restraints? The proper application of these can make a huge difference - and the defaults in phenix.refine could be better (and will be very soon). If you're willing to share the input files with us (off-list - [email protected] is the best way to reach all of the LBL developers), we can take a look and see if we notice anything. -Nat
Dear Bret, I'd bet your structure is pseudo-merohedrally twinned in P21 (twinning operator h,-k, -l). We had a similar problem with gp26 structure (an elongated fiber) that crystallized in a pseudo-P212121 cell, which, in reality was P21 with dimensions a = 40.5 b = 117.5 c =171.6 Å with the β-angle ~90.2° Refinement was stuck at ~32% before taking into account twinning. Good luck! Gino ****************************************************************************** Gino Cingolani, Ph.D. Associate Professor Thomas Jefferson University Dept. of Biochemistry & Molecular Biology 233 South 10th Street - Room 826 Philadelphia PA 19107 Office (215) 503 4573 Lab (215) 503 4595 Fax (215) 923 2117 E-mail: [email protected] Website: http://www.cingolanilab.org ****************************************************************************** "Nati non foste per viver come bruti, ma per seguir virtute e canoscenza" ("You were not born to live like brutes, but to follow virtue and knowledge") Dante, The Divine Comedy (Inferno, XXVI, vv. 119-120) From: [email protected] [[email protected]] on behalf of Bret Wallace [[email protected]] Sent: Sunday, March 18, 2012 11:37 AM To: [email protected] Subject: [phenixbb] Refinement using data with pseudo translational symmetry To all, First I apologize for the long email, but I wont to make sure that I give enough information to describe the problem. I have been working on a structure of two proteins in complex with each other (one forms a homodimer, with each monomer bound to 1 monomer of the other protein), using data, that according to phenix.xtriage contains pseudo translational symmetry (output pasted below). I have done a lot of searching of both the phenixBB and ccp4BB regarding solutions to this problem, but unfortunately most responses seem to be directed at resolving issues with MR. I have successfully performed MR using both phenix phaser, ccp4 phaser (the most updated version), as well as molrep. The issue arises upon structure refinement, where my Rwork and Rfree are essentially stuck at 28 and 33, respectively. I have built the entire structure, including ligands, except for any solvent molecules. Here are the details for the data/structure: SG: P212121 Cell: 72 124 175 90 90 90 Res: 50-2.8 Patterson analyses ------------------ Largest Patterson peak with length larger than 15 Angstrom Frac. coord. : 0.155 0.000 0.500 Distance to origin : 87.222 Height (origin=100) : 34.517 p_value(height) : 6.739e-04 The reported p_value has the following meaning: The probability that a peak of the specified height or larger is found in a Patterson function of a macro molecule that does not have any translational pseudo symmetry is equal to 6.739e-04. p_values smaller than 0.05 might indicate weak translational pseudo symmetry, or the self vector of a large anomalous scatterer such as Hg, whereas values smaller than 1e-3 are a very strong indication for the presence of translational pseudo symmetry. Xtriage notes that the if the PTS is crystallographic, that C2221 is a possible SG (x+1/6, y, z+1/2). However, neither XDS or HKL picks orthorhombic C, and the indexing/integrating doesn't work if I force it. I should also not there is no twinning detected by either xtriage or twinning servers. I have done a number of things to try to resolve this: -rescaling in lower symmetry (P21 and P1) -rescaling in the PG P222 and letting the MR program decide the proper space group -shifting origin and re-refining -a number of different refinement protocols (TLS, optimized weights/adp, simulated annealing, several cycles of rigid body, etc.) I have used HKL2000 and XDS to index/integrate/scale the data, both yield the same results, with slightly different completeness and Rmerge values, but refining with either gives similar R factor values. Does any know of any other possible things I should try to refine this data? I am happy to provide additional information upon request. Thanks in advance for any help! Bret
Nat, I have been using the auto.ncs function in phenix, but I haven't really looked to see if they are properly defined myself. Pavel, At this point I have not tried using the ordered_solvent command in phenix yet to find waters. I will try that real quick before I bother you to have a look at it. I'll let you know the updated R's afterwards. Gino, It is interesting that you say that, because I after I lowered the symmetry to P21, and re-ran xtriage, it mentioned that I had pseudo-merohedral symmetry, but I didn't pursue that further. My beta-angle was also ~90.2 in this case, and the twinning operator is the same (h, -k, -l). I will detwin and re-refine and see what happens, Thanks everyone and I'll let you know what works, Bret
On Sun, Mar 18, 2012 at 9:07 AM, Bret Wallace
I have been using the auto.ncs function in phenix, but I haven't really looked to see if they are properly defined myself.
The problem with the default NCS restraints is that they're too strict, and no amount of manual tweaking of atom selections can really overcome this. In this case I'd recommend trying two things: - use torsion NCS restraints (with XYZ weight optimization, probably): ncs.type=torsion, or equivalent menu in the GUI - turn off the restraints on NCS-related B-factors: ncs.b_factor_weight=0 You should probably get the latest nightly build for this. -Nat
Hi Bret,
I will detwin and re-refine and see what happens,
not sure what exactly you meant by this, but mentioning just in case: to use twinning information in refinement all you need is to give phenix.refine a twin law, like this: phenix.refine mode.pdb data.mtz twin_law="h,-k,-l" <other-parameters> (or whatever is equivalent the GUI). Pavel
Hi Bret, I think the apparent higher symmetry (P212121 vs P21) is due the (pseudo)-twinning operator (a 2-fold axis). So refinement with a twin law (h,-k,-l) will work only in the lower symmetry space group (P21). I hope it helps. Gino ****************************************************************************** Gino Cingolani, Ph.D. Associate Professor Thomas Jefferson University Dept. of Biochemistry & Molecular Biology 233 South 10th Street - Room 826 Philadelphia PA 19107 Office (215) 503 4573 Lab (215) 503 4595 Fax (215) 923 2117 E-mail: [email protected] Website: http://www.cingolanilab.org ****************************************************************************** "Nati non foste per viver come bruti, ma per seguir virtute e canoscenza" ("You were not born to live like brutes, but to follow virtue and knowledge") Dante, The Divine Comedy (Inferno, XXVI, vv. 119-120) ________________________________ From: [email protected] [[email protected]] on behalf of Bret Wallace [[email protected]] Sent: Sunday, March 18, 2012 12:07 PM To: PHENIX user mailing list Subject: Re: [phenixbb] Refinement using data with pseudo translational symmetry Nat, I have been using the auto.ncs function in phenix, but I haven't really looked to see if they are properly defined myself. Pavel, At this point I have not tried using the ordered_solvent command in phenix yet to find waters. I will try that real quick before I bother you to have a look at it. I'll let you know the updated R's afterwards. Gino, It is interesting that you say that, because I after I lowered the symmetry to P21, and re-ran xtriage, it mentioned that I had pseudo-merohedral symmetry, but I didn't pursue that further. My beta-angle was also ~90.2 in this case, and the twinning operator is the same (h, -k, -l). I will detwin and re-refine and see what happens, Thanks everyone and I'll let you know what works, Bret
Hi Bret, if you send me the model (the most current you have) and the data then I will have a look. Did you try to let phenix.refine build and refine water? Given the resolution and the fact that you did not add water the R-factors 28 and 33 seem reasonable (to me). Pavel On 3/18/12 8:37 AM, Bret Wallace wrote:
To all,
First I apologize for the long email, but I wont to make sure that I give enough information to describe the problem.
I have been working on a structure of two proteins in complex with each other (one forms a homodimer, with each monomer bound to 1 monomer of the other protein), using data, that according to phenix.xtriage contains pseudo translational symmetry (output pasted below). I have done a lot of searching of both the phenixBB and ccp4BB regarding solutions to this problem, but unfortunately most responses seem to be directed at resolving issues with MR. I have successfully performed MR using both phenix phaser, ccp4 phaser (the most updated version), as well as molrep. The issue arises upon structure refinement, where my Rwork and Rfree are essentially stuck at 28 and 33, respectively. I have built the entire structure, including ligands, except for any solvent molecules. Here are the details for the data/structure:
SG: P212121 Cell: 72 124 175 90 90 90 Res: 50-2.8
Patterson analyses ------------------
Largest Patterson peak with length larger than 15 Angstrom
Frac. coord. : 0.155 0.000 0.500 Distance to origin : 87.222 Height (origin=100) : 34.517 p_value(height) : 6.739e-04
The reported p_value has the following meaning: The probability that a peak of the specified height or larger is found in a Patterson function of a macro molecule that does not have any translational pseudo symmetry is equal to 6.739e-04. p_values smaller than 0.05 might indicate weak translational pseudo symmetry, or the self vector of a large anomalous scatterer such as Hg, whereas values smaller than 1e-3 are a very strong indication for the presence of translational pseudo symmetry.
Xtriage notes that the if the PTS is crystallographic, that C2221 is a possible SG (x+1/6, y, z+1/2). However, neither XDS or HKL picks orthorhombic C, and the indexing/integrating doesn't work if I force it.
I should also not there is no twinning detected by either xtriage or twinning servers.
I have done a number of things to try to resolve this:
-rescaling in lower symmetry (P21 and P1) -rescaling in the PG P222 and letting the MR program decide the proper space group -shifting origin and re-refining -a number of different refinement protocols (TLS, optimized weights/adp, simulated annealing, several cycles of rigid body, etc.)
I have used HKL2000 and XDS to index/integrate/scale the data, both yield the same results, with slightly different completeness and Rmerge values, but refining with either gives similar R factor values.
Does any know of any other possible things I should try to refine this data? I am happy to provide additional information upon request.
Thanks in advance for any help! Bret
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
participants (4)
-
Bret Wallace -
Gino Cingolani -
Nathaniel Echols -
Pavel Afonine