March 2007 - phenixbb - phenix-online.org

Re: [phenixbb] coordination of metal ions
by Ralf W. Grosse-Kunstleve 29 Mar '07

29 Mar '07

Hi Jesper, could you send me the fragments from your PDB file with the metal ion and the bound residues? I'll work this into an example for making the required cif definitions manually. Ralf

1 0

coordination of metal ions
by Jesper Lykkegaard Karlsen 29 Mar '07

29 Mar '07

Dear Phenix-devels Is it possible set up geometric restrains for metal ions bound to protein in phenix.refine? elbow.builder seems only to work on separate molecules and not on metal ions with its coordinating sidechains. I found the feature to introduce custom bond distance between two atoms, but what about custom angles? And by the way will this these custom setting ignore Van Der Waals radii? or is there a way to set these for special atoms? Thanks! Jesper

2 1

riding hydrogen and mlhl
by Jianghai Zhu 28 Mar '07

28 Mar '07

Hi All, I just found a refinement.hydrogens entry in the phenix.refine. Can we use phenix.refine to generate riding hydrogens to help the geometry in low- or mid-resolution refinement now? Using hydrogens in refmac5 really gives a better result than not using hydrogens. Another question. Is there a way to blur the experimental or density modified phases when using mlhl target? Thanks. Jianghai +++++++++++++++++++++++++++++++ Jianghai Zhu, Ph.D CBR Institute for Biomedical Research Department of Pathology Harvard Medical School 200 Longwood Ave., Boston, MA 02115 Ph: 617-278-3211 Fx: 618-278-3232 +++++++++++++++++++++++++++++++

3 4

NCS weighting
by Jianghai Zhu 28 Mar '07

28 Mar '07

Hi, I am wondering how the NCS weighting in phenix.refine works. I guess the default values (coordinate_sigma = 0.05, b_factor_weight = 10) are pretty tight. What kind values would be appropriate for medium and loose NCS? Jianghai +++++++++++++++++++++++++++++++ Jianghai Zhu, Ph.D CBR Institute for Biomedical Research Department of Pathology Harvard Medical School 200 Longwood Ave., Boston, MA 02115 Ph: 617-278-3211 Fx: 618-278-3232 +++++++++++++++++++++++++++++++

4 6

Re: [phenixbb] ls_wunit_k1
by Ralf W. Grosse-Kunstleve 26 Mar '07

26 Mar '07

> All right. The target for my TLS is also LS. Is that intentional too? Pavel, who wrote the TLS code, is current offline. I don't know if Pavel ran systematic tests to determine the default target for TLS. A couple of months ago I ran systematic tests exercising rigid body refinement using both the LS and ML targets. Overall it turned out that ML is slightly better than LS. For an individual structure and a particular starting point it can be the other way around, but averaged over about 70 structures with 100 random starting points each ML generally outperforms LS. We should have updated the default accordingly, but it fell through the cracks. If you have the time to try out both ML and LS in the refinement of your structure, we'd be interested to know the results. Ralf P.S.: What's the version of phenix.refine you are using? -- I'm still puzzled why you didn't get the "automatic adjustment" messages.

4 5

Re: [phenixbb] ls_wunit_k1
by Ralf W. Grosse-Kunstleve 26 Mar '07

26 Mar '07

> You are right. Somehow, the default target function in > refinement.bulk_solvent_and_scale is LS, not ML. The default (phenix.refine --show-defaults=all) is actually ml. However, if the main target is set to ls the bulk_solvent_and_scale target is automatically reset to ls, too. There should be corresponding messages in the log file ("automatic adjustment"). Ralf

3 5

ls_wunit_k1
by Jianghai Zhu 26 Mar '07

26 Mar '07

Hi, I happen to notice that the target function of "Bulk solvent correction and scaling" in a phenix.refine run is ls_wunit_k1. Does that mean LS function for bulk solvent correction and scaling is the default in phenix.refine? If I uncheck ls_wunit_k1 in refinement.ls_target_names in my parameter file, will phenix.refine use the ML function or something bad will happen? Jianghai +++++++++++++++++++++++++++++++ Jianghai Zhu, Ph.D CBR Institute for Biomedical Research Department of Pathology Harvard Medical School 200 Longwood Ave., Boston, MA 02115 Ph: 617-278-3211 Fx: 618-278-3232 +++++++++++++++++++++++++++++++

2 3

Re: [phenixbb] The performance of phenix.refine
by Ralf W. Grosse-Kunstleve 22 Mar '07

22 Mar '07

> Did anybody actually compare the performance among phenix.refine, CNS > and refmac5? I am trying the phenix.refine from the latest CCI > bundle for my refinement. I have a relative large structure (800 kD > unique) and unit cell (140, 160, 550, 90, 90, 90). The data is at > low resolution ( 3.8 A) with experimental phases. My impression is > that phenix.refine takes a much longer time to finish the jobs, say > rigid body refinement or SA refinement, and it also takes a lot more > memories ( in my case, phenix.refine uses more than 700 MB, but other > programs use less than 200 MB). We have been concentrating on features and a high level of automation much more than raw speed and memory efficiency. Given that memory is so cheap these days (ca. $80/GB) there isn't much incentive for us to trade developer time for a smaller memory footprint. Speed optimizations are a bit higher up on the list. > Is this because the python code in > phenix.refine? CNS and refmac5 are written in FROTRAN, I think. Just > curious. phenix.refine doesn't spend much time in the Python code. It is a long while ago I did careful timings, but I'd guess it is significantly less than 10% of the time. C++ is intrinsically almost as fast as FORTRAN if the same development time is invested in the code. I.e. what you see is not a language issue, just a tradeoff features vs. optimization. We'll work on making phenix.refine a bit faster in the future, but at the minute more features are still our focus. Ralf

1 0

The performance of phenix.refine
by Jianghai Zhu 22 Mar '07

22 Mar '07

Hi, Did anybody actually compare the performance among phenix.refine, CNS and refmac5? I am trying the phenix.refine from the latest CCI bundle for my refinement. I have a relative large structure (800 kD unique) and unit cell (140, 160, 550, 90, 90, 90). The data is at low resolution ( 3.8 A) with experimental phases. My impression is that phenix.refine takes a much longer time to finish the jobs, say rigid body refinement or SA refinement, and it also takes a lot more memories ( in my case, phenix.refine uses more than 700 MB, but other programs use less than 200 MB). Is this because the python code in phenix.refine? CNS and refmac5 are written in FROTRAN, I think. Just curious. Jianghai +++++++++++++++++++++++++++++++ Jianghai Zhu, Ph.D CBR Institute for Biomedical Research Department of Pathology Harvard Medical School 200 Longwood Ave., Boston, MA 02115 Ph: 617-278-3211 Fx: 618-278-3232 +++++++++++++++++++++++++++++++

1 0

Re: [phenixbb] freeR syntax question in phenix.refine
by Ralf W. Grosse-Kunstleve 20 Mar '07

20 Mar '07

> I was running refinement from an mtz file from sigmaa in ccp4 with an > appropriate FreeR flag designated in the MTZ file. phenix.refine > worked fine and seemed to exclude the test reflections, but I wanted > to know what these commands do. Don't! :-) > r_free_flags { > file_name = "S19_sigmaa1.mtz" > label = "FreeR_flag" > test_flag_value = 0 --> don't know what this means? > disable_suitability_test = False --> implications of this? phenix.refine is designed to figure everything out automatically. You will have to specify file_name and label only if you have multiple R-free-flag arrays in the reflection files you give to phenix.refine, which is very unusual. phenix.refine does some sanity checks to determine if arrays that "look like" R-free-flag arrays are actually suitable for refinement, and reject them if "it doesn't think so". It will then tell you that you can override the sanity checks by setting disable_suitability_test to True. Again, this is very unusual. The test_flag_value could potentially be important. phenix.refine analyses the putative R-free-flag array(s) to guess if they came from CNS, where 0 usually means work set, 1 free set, or CCP4, which mostly uses values from 0 to 19, 0 is the default free set in REFMAC. Therefore, roughly, if phenix.refine "thinks" it is a CNS R-free-flag array it assigns test_flag_value=1, or 0 if it "thinks" it is a CCP4 array. Only if you used REFMAC before for refinement, and only if you manually instructed REFMAC to use a value different from 0 for the free set, will you have to manually tell phenix.refine about it, e.g. via r_free_flags.test_flag_value=19. If you are interested in more details, inspect the script iotbx/iotbx/reflection_file_utils.py, which does all the guesswork. Look for "get_r_free_flags_scores".

1 0