Good point. For now the best thing to do is to read the RvsR paper cited in the text. I'll add a section to the manual as well. Ideally of course, xtriage should make its own judgement. Some code is there, but I took it out as it is not well-tested (and didn't do the job) P 2009/8/25 Pavel Afonine :

Hi Peter,

do you have any guidance to how interpret this table summarized somewhere in the documentation? Otherwise it looks a bit cryptic to me (I tried it while ago, may be it is improved now). Alternatively, it would be nice if Xtriage prints out its own verdict based on that table.

Thanks! Pavel.

On 8/25/09 10:47 AM, Peter Zwart wrote:

...

Subsequently, run

phenix.xtriage p1data.mtz reference.structure.file=MR.1.pdb

and start interpreting the RvsR tables.

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb

-- ----------------------------------------------------------------- P.H. Zwart Beamline Scientist Berkeley Center for Structural Biology Lawrence Berkeley National Laboratories 1 Cyclotron Road, Berkeley, CA-94703, USA Cell: 510 289 9246 BCSB: http://bcsb.als.lbl.gov PHENIX: http://www.phenix-online.org CCTBX: http://cctbx.sf.net -----------------------------------------------------------------

Hi Maia, Two points: 1) : Given that fact that you can process the data in p622, you can expand the data to p1: phenix.reflection_file_converter mydata.sca --sca=p1.sca --expand_to_p1 Eventhough this is not the same as having 'real' p1 data, you know that the data merges well in p622, so no harm is done (or at least not a lot). Do you have pseudo translational symmetry by any chance? I wasn't able to find any relation between the two provided unit cells. It is very curious that their volume are equal though. 2): The fact that you cannot solve it might also be due to the fact that your MR model is incorrect for some obscure reason. HTH Peter 2009/8/25 Maia Cherney :

Hi everybody,

I have two types of hexagonal crystals: rods and bipyramids with dimensions 150x150x83 (rods) and 142x142x93 (bipyramids). Both types of crystals give the same spacegroup P6222. However, the rods gave a good solution and refined to low R factors, whereas the bipyramids give a solution with high LLG and Z-score, but don't refine at all (R factors around 50%). The xtriage does not indicate any twinning. Lower symmetry spacegroups (trigonal and monoclinic) have the same problem: give solutions with high scores that would not refine. The resolution is 2.3A. Unfortunately, the dataset processed in p1 has only 60% completeness (as it was collected according to p3 strategy). Should I try to solve in p1 with this low completeness? What  can be a problem?

Maia

Peter Zwart wrote:

...

Good point.

For now the best thing to do is to read the RvsR paper cited in the text. I'll add a section to the manual as well.

Ideally of course, xtriage should make its own judgement. Some code is there, but I took it out as it is not well-tested (and didn't do the job)

P

2009/8/25 Pavel Afonine :

...

Hi Peter,

do you have any guidance to how interpret this table summarized somewhere in the documentation? Otherwise it looks a bit cryptic to me (I tried it while ago, may be it is improved now). Alternatively, it would be nice if Xtriage prints out its own verdict based on that table.

Thanks! Pavel.

On 8/25/09 10:47 AM, Peter Zwart wrote:

...

Subsequently, run

phenix.xtriage p1data.mtz reference.structure.file=MR.1.pdb

and start interpreting the RvsR tables.

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb

-- ----------------------------------------------------------------- P.H. Zwart Beamline Scientist Berkeley Center for Structural Biology Lawrence Berkeley National Laboratories 1 Cyclotron Road, Berkeley, CA-94703, USA Cell: 510 289 9246 BCSB: http://bcsb.als.lbl.gov PHENIX: http://www.phenix-online.org CCTBX: http://cctbx.sf.net -----------------------------------------------------------------

Hi Peter, how did you generate that plot? Is it a general app, or by hand? Because it's a cool way for getting your head around the twinning problems... Cheers phx Peter Zwart wrote:

Can you send the exact unit cell parameters? The number you give now are to rough to make a fair comparison.

In any case, C2221 does lie directly below P43212 (see attached figure), so you have a good chance that is a possibility. On the other hand, so are P43 and P212121. It is fairly straightforwadr to try all, just cumbersome. I suggest however you get your data in P1, and solve it in that spacegroup. Subsequently, run

phenix.xtriage p1data.mtz reference.structure.file=MR.1.pdb

and start interpreting the RvsR tables. Contact me directly if you have difficulties with this, or share with the bb as this is a very instructive exercise.

Cheers

Peter

2009/8/25 Leigh Allen :

...

Hello Phenix users.

I recently solved the structure of my enzyme using SAD to a resolution of 1.9A. The R values were 0.19/0.25 and the space group was P43212 with a unit cell ~80, 80, 220.

I then starting collecting datasets cocrystallized with ATP analogs and noticed that a lot of my datasets have a different unit cell when scaled in P43212. It comes out to be ~60, 60, 220 and I can NOT get a solution in Phaser with this data. If I scale it in C2221, then the unit cell is the same (80, 80, 220), but upon refinement, the R factors don't go much below 0.26/0.32. Has anyone ever seen something like this happen? I'm unsure if I should feel like the solution is C2221 since there are space groups of higher symmetry with similar residuals that seem plausible in indexing just the smaller unit cell. Should I keep trying to get a solution out of the data that leads to different unit cell parameters, but has higher symmetry?

Xtriage seems to find twinning, but I think it's from NCS since my ASU contains a homodimer. I could also be wrong about this...

Thanks in advance for any advice!

******* Leigh Allen Ph.D Candidate McCafferty Lab Duke University Department of Chemistry

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb

------------------------------------------------------------------------

------------------------------------------------------------------------

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb

Ah, Peter, cool stuff indeed. I tried to follow your procedure with P65: phenix.python $somedriectory/cctbx/examples/sub_group_graphy.py P66 > tmp.com chmod +x tmp.com tmp.com # fails: bash: tmp.com: command not found ./tmp.com # also fails with the following error message ./tmp.com: line 2: ----------------------------: command not found ./tmp.com: line 4: Outgoing: command not found ./tmp.com: line 5: P: command not found ./tmp.com: line 6: P: command not found ./tmp.com: line 7: P: command not found ./tmp.com: line 8: P: command not found ./tmp.com: line 10: syntax error near unexpected token `(' ./tmp.com: line 10: `P 1 ---> P 65 :: using: (x-y,x,z+5/6) (y,-x+y,z+1/6) (-y,x-y,z+2/3) (-x+y,-x,z+1/3) (-x,-y,z+1/2) symops left: 0' Three empty files are generated: [], [P 65, and [P 65]. Any suggestions? Dot is installed. OS is RHEL 5.4. Phenix is 1.4-3 (cctbx tag: 2008_12_07_1353) Andreas On Wed, Aug 26, 2009 at 11:36 PM, Peter Zwart wrote:

...

 Because it's a cool way for getting your head around the twinning problems... Cheers phx

Indeed, that is what it was written for.

try something like

phenix.python $somedriectory/cctbx/examples/sub_group_graphy.py  P432 > tmp.com chmod +x tmp.com tmp.com display sg_graph.png

you need to have dot installed (comes with ccp4)

HTH

Peter

...

Peter Zwart wrote:

...

Can you send the exact unit cell parameters? The number you give now are to rough to make a fair comparison.

In any case, C2221 does lie directly below P43212 (see attached figure), so you have a good chance that is a possibility. On the other hand, so are P43 and P212121. It is fairly straightforwadr to try all, just cumbersome. I suggest however you get your data in P1, and solve it in that spacegroup. Subsequently, run

phenix.xtriage p1data.mtz reference.structure.file=MR.1.pdb

and start interpreting the RvsR tables. Contact me directly if you have difficulties with this, or share with the bb as this is a very instructive exercise.

Cheers

Peter

2009/8/25 Leigh Allen :

...

Hello Phenix users.

I recently solved the structure of my enzyme using SAD to a resolution of 1.9A.  The R values were 0.19/0.25 and the space group was P43212 with a unit cell ~80, 80, 220.

I then starting collecting datasets cocrystallized with ATP analogs and noticed that a lot of my datasets have a different unit cell when scaled in P43212.  It comes out to be ~60, 60, 220 and I can NOT get a solution in Phaser with this data.  If I scale it in C2221, then the unit cell is the same (80, 80, 220), but upon refinement, the R factors don't go much below 0.26/0.32.  Has anyone ever seen something like this happen?  I'm unsure if I should feel like the solution is C2221 since there are space groups of higher symmetry with similar residuals that seem plausible in indexing just the smaller unit cell. Should I keep trying to get a solution out of the data that leads to different unit cell parameters, but has higher symmetry?

Xtriage seems to find twinning, but I think it's from NCS since my ASU contains a homodimer.  I could also be wrong about this...

Thanks in advance for any advice!

******* Leigh Allen Ph.D Candidate McCafferty Lab Duke University Department of Chemistry

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb

------------------------------------------------------------------------

------------------------------------------------------------------------

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb

-- ----------------------------------------------------------------- P.H. Zwart Beamline Scientist Berkeley Center for Structural Biology Lawrence Berkeley National Laboratories 1 Cyclotron Road, Berkeley, CA-94703, USA Cell: 510 289 9246 BCSB:     http://bcsb.als.lbl.gov PHENIX: http://www.phenix-online.org CCTBX:  http://cctbx.sf.net -----------------------------------------------------------------

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb

Thanks, Folmer, that did the trick. So obvious, I should have tried it myself... Andreas On Fri, Aug 28, 2009 at 1:05 PM, Folmer Fredslund wrote:

Dear Andreas

2009/8/28 Andreas Forster :

...

Ah, Peter, cool stuff indeed.  I tried to follow your procedure with P65:

phenix.python $somedriectory/cctbx/examples/sub_group_graphy.py  P66 > tmp.com chmod +x tmp.com tmp.com  # fails:  bash:  tmp.com:  command not found ./tmp.com # also fails with the following error message

./tmp.com: line 2: ----------------------------: command not found ./tmp.com: line 4: Outgoing: command not found ./tmp.com: line 5: P: command not found ./tmp.com: line 6: P: command not found ./tmp.com: line 7: P: command not found ./tmp.com: line 8: P: command not found ./tmp.com: line 10: syntax error near unexpected token `(' ./tmp.com: line 10: `P 1  --->  P 65  ::    using: (x-y,x,z+5/6) (y,-x+y,z+1/6)  (-y,x-y,z+2/3)  (-x+y,-x,z+1/3)  (-x,-y,z+1/2) symops left: 0'

Three empty files are generated: [], [P 65, and [P 65].

Any suggestions?  Dot is installed.  OS is RHEL 5.4.  Phenix is 1.4-3 (cctbx tag: 2008_12_07_1353)

I just checked this myself and what I found worked perfectly was to actually read some of the output that are in the tmp.com file.

I get something like the following in the end of the tmp.com file:

dot -Tpng > sg_graph.png << EOF  digraph f { rankdir=LR"P 1" -> "P 65" ;"P 1" -> "P 32" ;"P 1" -> "P 1 1 21" ;"P 1 1 21" -> "P 65" ;"P 32" -> "P 65" ;} EOF

Simply pasting this in a terminal seems to work perfectly.

Best regards, Folmer Fredslund

...

Andreas

On Wed, Aug 26, 2009 at 11:36 PM, Peter Zwart wrote:

...

...

 Because it's a cool way for getting your head around the twinning problems... Cheers phx

Indeed, that is what it was written for.

try something like

phenix.python $somedriectory/cctbx/examples/sub_group_graphy.py  P432 > tmp.com chmod +x tmp.com tmp.com display sg_graph.png

you need to have dot installed (comes with ccp4)

HTH

Peter

...

Peter Zwart wrote:

...

Can you send the exact unit cell parameters? The number you give now are to rough to make a fair comparison.

In any case, C2221 does lie directly below P43212 (see attached figure), so you have a good chance that is a possibility. On the other hand, so are P43 and P212121. It is fairly straightforwadr to try all, just cumbersome. I suggest however you get your data in P1, and solve it in that spacegroup. Subsequently, run

phenix.xtriage p1data.mtz reference.structure.file=MR.1.pdb

and start interpreting the RvsR tables. Contact me directly if you have difficulties with this, or share with the bb as this is a very instructive exercise.

Cheers

Peter

2009/8/25 Leigh Allen :

...

Hello Phenix users.

I recently solved the structure of my enzyme using SAD to a resolution of 1.9A.  The R values were 0.19/0.25 and the space group was P43212 with a unit cell ~80, 80, 220.

I then starting collecting datasets cocrystallized with ATP analogs and noticed that a lot of my datasets have a different unit cell when scaled in P43212.  It comes out to be ~60, 60, 220 and I can NOT get a solution in Phaser with this data.  If I scale it in C2221, then the unit cell is the same (80, 80, 220), but upon refinement, the R factors don't go much below 0.26/0.32.  Has anyone ever seen something like this happen?  I'm unsure if I should feel like the solution is C2221 since there are space groups of higher symmetry with similar residuals that seem plausible in indexing just the smaller unit cell. Should I keep trying to get a solution out of the data that leads to different unit cell parameters, but has higher symmetry?

Xtriage seems to find twinning, but I think it's from NCS since my ASU contains a homodimer.  I could also be wrong about this...

Thanks in advance for any advice!

******* Leigh Allen Ph.D Candidate McCafferty Lab Duke University Department of Chemistry

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb

------------------------------------------------------------------------

------------------------------------------------------------------------

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb

-- ----------------------------------------------------------------- P.H. Zwart Beamline Scientist Berkeley Center for Structural Biology Lawrence Berkeley National Laboratories 1 Cyclotron Road, Berkeley, CA-94703, USA Cell: 510 289 9246 BCSB:     http://bcsb.als.lbl.gov PHENIX: http://www.phenix-online.org CCTBX:  http://cctbx.sf.net -----------------------------------------------------------------

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb

_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb