- phenixbb - phenix-online.org

Phenix.Refine: Missing High Resolution Data
by Mark A. White 16 Mar '18

16 Mar '18

Hello, I have several data sets which seem to loose completeness in their high-resolution shells when the HKL *.sca file is converted into an MTZ. I am using the recommended Diederichs CC1/2 cut off of 0.5, which is usually an I/s <2. I also have an usually high redundancy in these data sets, not sure if that has any effect. Here is a typical comparison of the Phenix statistics and the HKL values. Note that the resolution bins do not match, and that the Phenix refinement was truncated to 3.0Å. PHENIX HKL2000/Scalepack RESOLUTION RANGE COMPL total CC1/2 CC* Redn ScP-Angstrom 50.1353 - 8.8720 97 97.8 0.989 0.997 10.9 70.00 7.87 8.8720 - 7.0483 99 - 7.0483 - 6.1591 100 99.6 0.990 0.997 12.8 7.87 6.25 6.1591 - 5.5968 100 99.7 0.991 0.998 12.9 6.25 5.46 5.5968 - 5.1961 99 - 5.1961 - 4.8900 99 99.6 0.991 0.998 13.0 5.46 4.96 4.8900 - 4.6453 99 99.5 0.986 0.996 12.9 4.96 4.60 4.6453 - 4.4432 99 99.2 0.990 0.997 13.1 4.60 4.33 4.4432 - 4.2723 99 - 4.2723 - 4.1249 99 99.8 0.988 0.997 13.3 4.33 4.11 4.1249 - 3.9960 99 99.9 0.989 0.997 13.4 4.11 3.94 3.9960 - 3.8818 99 99.9 0.987 0.997 13.5 3.94 3.78 3.8818 - 3.7796 100 - 3.7796 - 3.6875 100 99.8 0.985 0.996 13.4 3.78 3.65 3.6875 - 3.6037 100 100.0 0.975 0.994 13.5 3.65 3.54 3.6037 - 3.5270 100 - 3.5270 - 3.4565 100 100.0 0.964 0.991 13.5 3.54 3.44 3.4565 - 3.3912 100 - 3.3912 - 3.3307 97 99.9 0.966 0.991 12.1 3.44 3.35 3.3307 - 3.2742 87 100.0 0.916 0.978 10.2 3.35 3.27 3.2742 - 3.2214 68 100.0 0.857 0.961 8.5 3.27 3.19 3.2214 - 3.1719 54 - 3.1719 - 3.1252 41 100.0 0.763 0.930 7.4 3.19 3.12 3.1252 - 3.0812 33 99.9 0.616 0.873 6.4 3.12 3.06 3.0812 - 3.0396 24 - 3.0396 - 3.0001 18 98.4 0.547 0.841 5.8 3.06 3.00 95.1 0.307 0.685 5.2 3.00 2.95 89.8 0.302 0.681 4.8 2.95 2.90 98.9 0.982 0.995 10.9 All reflect. -- Yours sincerely, Mark A. White, Ph.D. Associate Professor of Biochemistry and Molecular Biology, Manager, Sealy Center for Structural Biology and Molecular Biophysics Macromolecular X-ray Laboratory, Basic Science Building, Room 6.658A University of Texas Medical Branch Galveston, TX 77555-0647 Tel. (409) 747-4747 mailto://[email protected] http://xray.utmb.edu QQ: "Don’t persist in folly. Some people commit themselves to their errors. They commit one error and consider it constancy to go on that way." - Baltasar Gracian (1658)

4 4

cis-Pro and trans-Pro refinement
by 张士军 16 Mar '18

16 Mar '18

Dear all I am refining a structure which has cis-Pro and trans-Pro, the tans-Pro is gone when I set the "threshold degrees for cis-peptide " from default 45 to 65, but still has cis-Pro. While no significant change when I set it to 15. My question is how to set in phenix refinement in phenix GUI to clear the Pro residues in cis- or trans- conformations. Best Regards shijun

2 1

Phenix refine ligand with alt conf and occupancy<1
by Leonhard Kick 16 Mar '18

16 Mar '18

Dear all, I'm currently working on a structure with a series of ligands which have occupancies of ~80% (according to phenix.refine). So far so good, a few of these also show alternate conformations which I'd like to let phenix refine as well, but phenix.refine adjusts the occupancy of the whole ligand (conformation A+B) automatically to 1. I see that this is useful in most cases, but not in this one. Is there a possibility to adjust the overall occupancy or another work-around that anyone can suggest? Thank you Best Leo -- ================================================== Leonhard Kick AG Schneider Technische Universität München Dept. Chemistry, Chair of Biochemistry Lichtenbergstr. 4 D-85748 Garching b. München +49 89 289 13359 leonhard.kick(a)tum.de ==================================================

2 1

Real space refinement of complete virus structure
by #TOFAYEL AHMED# 13 Mar '18

13 Mar '18

Dear all, I am trying to refine a complete virus pdb structure (possessing icosahedral symmetry) in a cryo-EM map using phenix.real_space_refine. The run fails complaining "Multi-model PDB (with MODEL-ENDMDL) is not supported". So, I followed a previous post in phenixbb: http://www.phenix-online.org/pipermail/phenixbb/2011-May/017079.html and merged multiple models into one but faced following problems: 1. This merged file is unreadable in UCSF Chimera. 2. The secondary structure definition in the PDB file header is not recognizable anymore in Phenix. In my understanding, both the problems are arising due to PDB format's inability to handle more than 62 chains and 99999 atoms in a file. So, what will be the work-around to perform a real space refinement of the complete virus structure at once? Also, how to maintain secondary structure during such refinement? Any help is highly appreciated. Best regards, Tofayel ............................................ Tofayel Ahmed Graduate Student (PhD) | Lab-02n-01 School of Biological Sciences Nanyang Technological University 60 Nanyang Drive, Singapore 637551 +65 8650.0473 |tofayel001(a)e.ntu.edu.sg<mailto:[email protected]>

2 1

AutoSol Br SAD/MAD
by Ian Slaymaker 12 Mar '18

12 Mar '18

Hello Phenixbb, I've been out of the crystallography game for a few years and was hoping for some suggestions. I'm trying to phase a structure (~1200aa protein:DNA complex) with native data out to 2.0A and SAD/MAD data out to 3.0-2.6. I'm trying to use Br/I soaking while I wait for Sel-Met crystals to grow. I'm collecting 1800 frames at a synchrotron and scaling together multiple MAD wavelength sets (30 degree inverse wedges) collected from the same crystal with phenix.scale_and_merge to achieve >85% anomalous data (determined by xtriage). Anomalous signal of the merged data is around 10 at the 3.5 angstrom shell. I'm keeping an eye on the merging statistics, but I'm not sure what the critical stats to monitor are. AutoSol has given me a a few possible solutions, finding between 3 and 8 sites for MAD and SAD sets(I'm letting AutoSol decide how many to find) with Resolve Rfactors around 0.34, but low map skew (<1 and sometimes negative). I've tried using the initial sites to search for additional Br sites, but with no luck, and the density isn't good enough to build anything into. I'm doing hyss searches with various resolution cutoffs and continue to get the same/similar uninterpretable density with very few heavy atom sites found. My questions: What are the critical stats to look at when merging data for anomalous signal? Are there parameters to explore in AutoSol to identify more heavy atom sites? Should I accept that I need to wait for new crystals to grow or keep charging ahead with this data, encouraged by some promising numbers by AutoSol? Thank you in advance for the suggestions. -Ian

2 1

View Solvent Mask
by ben flath 09 Mar '18

09 Mar '18

Hi All I was watching the YouTube tutorial on Polder Maps and noticed that the bulk solvent mask was visualized in coot. How can I visualize my bulk solvent mask? Regards Ben Flath

4 5

Re: [phenixbb] View Solvent Mask
by Flath, Ben 09 Mar '18

09 Mar '18

It seems I am looking for the the compute_box option in polder maps. Ben Sent from my Android -------- Original message -------- From: "Flath, Ben" <bef206(a)mail.usask.ca> Date: 2018-03-09 11:15 AM (GMT-06:00) To: Pavel Afonine <pafonine(a)lbl.gov>, "Flath, Ben" <bef206(a)mail.usask.ca>, phenixbb(a)phenix-online.org Subject: Re: [phenixbb] View Solvent Mask Now the important question How can I modify the mask so it doesn't enter pockets within my model? Add HOH with occupancy set to zero and ignore_zero_occupancy_atoms=False? Or is running polder maps more effective? Thanks again Ben Sent from my Android -------- Original message -------- From: Pavel Afonine <pafonine(a)lbl.gov> Date: 2018-03-09 10:50 AM (GMT-06:00) To: "Flath, Ben" <bef206(a)mail.usask.ca>, phenixbb(a)phenix-online.org Subject: Re: [phenixbb] View Solvent Mask Hi Ben, I was watching the YouTube tutorial on Polder Maps and noticed that the bulk solvent mask was visualized in coot. How can I visualize my bulk solvent mask? phenix.mask model.pdb will calculate the mask. Pavel

1 0

Questions about the Autosol water placing option
by Kaibo Zhang 07 Mar '18

07 Mar '18

Helllo, I was using AutoSol GUI for MR-SAD of my data and trying to turn off "place waters in refinement". However, the output pdb files still got waters in them. I did check the run log file and confirmed both the "place_waters" and the "flood_with_waters" flags were set to false. But in the later session, it still recorded "adding waters in refinement". Is there any parameters I should adjust to avoid adding waters during the Autosol run. Or is it just turned on in the end in order to calculate R_work/free, and doesn't affect your phasing solution? Look forward to your responses. Thank you in advance! Regards Kaibo *Kaibo* PULSe, Purdue University Hockmeyer Hall, Room 321 240 S. Martin Jischke Drive, West Lafayette, IN 47907 Telephone: 765-337-1955 E-mail: zhang644(a)purdue.edu

2 1

Re: [phenixbb] Overlapping ligand-protein side chain density
by van den Bedem, Henry 02 Mar '18

02 Mar '18

If the phe is a gatekeeper, shouldn’t the ligand be refined at partial occupancy; i.e. occupancies not necessarily have to ‘add up’ as you suggest? Maybe this link is helpful too: www.biorxiv.org/content/early/2018/01/24/253419 H From: <phenixbb-bounces(a)phenix-online.org> on behalf of "Phan, Jason" <jason.phan(a)vanderbilt.edu> Date: Wednesday, February 28, 2018 at 1:30 PM To: "phenixbb(a)phenix-online.org" <phenixbb(a)phenix-online.org> Subject: [phenixbb] Overlapping ligand-protein side chain density All, A piece of a ligand and the side chain of a “gate-keeper” Phe occupy the same space with well-defined features observed for both (resolution is 1.7 A). It looks about 50:50. How do you refine both ligand and protein side chain in this case? A couple of phenixbb suggestions for dealing with ligand-ligand overlapping density have been considered. With the first suggestion, the two entities still clashed and moved apart off of their respective densities. The second suggestion is not applicable in this case since the other molecule is not a ligand but part of the protein but it was tested anyway. Although there was no bumping, the occupancies don’t add up, resulting in a big blob of negative density around the ligand piece. ------------------------------ Only two comments: 1. At that resolution, constrained group occupancy refinement should work reasonably well (provided you can model the 2 entities). Then you also do not have clashes between the molecules, because Occ(A)+Occ(B)=1, meaning when one (A) is there, the other one (B) is not. This works with refmac (external keyword file); if you need more sophisticated occupancy re/constraints SHELXL may offer more opportunities. 2. There is no necessity for the two NCS copies of the binding site to look exactly the same (non-equivalent). Maybe there is a good reason/story (accessibility, contacts etc) for one site to be occupied differently than the other one. Best, BR —————————————— Modeling two molecules that occupy overlapping binding sites in a structure simply involves designating them as alternate conformers, with the same chain and residue number, and an occupancy that sums to 1.0. For example, if you have an AMP and an ADP that occupy the same binding site, you would define them as AAMP B 501 BADP B 501 and initially set the occupancies for the atoms in each conformer to the ratio (50:50, 30:70, etc.) that you observe in the density. Refinement in this manner is straightforward in PHENIX. Diana ----------------------------

5 6

Overlapping ligand-protein side chain density
by Phan, Jason 28 Feb '18

28 Feb '18

All, A piece of a ligand and the side chain of a “gate-keeper” Phe occupy the same space with well-defined features observed for both (resolution is 1.7 A). It looks about 50:50. How do you refine both ligand and protein side chain in this case? A couple of phenixbb suggestions for dealing with ligand-ligand overlapping density have been considered. With the first suggestion, the two entities still clashed and moved apart off of their respective densities. The second suggestion is not applicable in this case since the other molecule is not a ligand but part of the protein but it was tested anyway. Although there was no bumping, the occupancies don’t add up, resulting in a big blob of negative density around the ligand piece. ------------------------------ Only two comments: 1. At that resolution, constrained group occupancy refinement should work reasonably well (provided you can model the 2 entities). Then you also do not have clashes between the molecules, because Occ(A)+Occ(B)=1, meaning when one (A) is there, the other one (B) is not. This works with refmac (external keyword file); if you need more sophisticated occupancy re/constraints SHELXL may offer more opportunities. 2. There is no necessity for the two NCS copies of the binding site to look exactly the same (non-equivalent). Maybe there is a good reason/story (accessibility, contacts etc) for one site to be occupied differently than the other one. Best, BR —————————————— Modeling two molecules that occupy overlapping binding sites in a structure simply involves designating them as alternate conformers, with the same chain and residue number, and an occupancy that sums to 1.0. For example, if you have an AMP and an ADP that occupy the same binding site, you would define them as AAMP B 501 BADP B 501 and initially set the occupancies for the atoms in each conformer to the ratio (50:50, 30:70, etc.) that you observe in the density. Refinement in this manner is straightforward in PHENIX. Diana ----------------------------

2 1