rfree checksums version 1.3 and 1.4
Hi am using the same exact mtz data file from CCP4 uniquefy for two refinements in phenix. I did one round of refinement in phenix 1.3 and then used the output pdb of that refinement and input into the next round of refinement . Only I switched versions to phenix version 1.4. Most other things weere the same including the resolution cutoff used and the input mtz. But phenix tells me that the md5checksums dont match. i am very sure its the same input mtz and identical data resolution ( both cases specifying refinement.input.xray_data.high_resolution=3.0) Could this message of mismatched checksums be an error due to chaging versions midtsream . I certainly dont want to be trampling on my Rfree reflections but apart from copying files to a new directory for the next round and changing phenix versions I have changed little else Thanks for your help in advance hari
Most other things weere the same including the resolution cutoff used and the input mtz. But phenix tells me that the md5checksums dont match.
Sorry, that's unexpected. The cutoff doesn't matter. The checksum is computed on the input data, before the cutoff is applied. The checksum is just a safety guard to avoid accidents. If you are sure the test set has not changed, simply remove the remark with the checksum from the pdb file. Let me know if the problem happens again and I'll take a closer look. In that case I will need all the input files and the sequence of commands to reproduce the problem (including the command that wrote the pdb file with the checksum). Ralf
Hello Ralf and eversone else
Thanks for your email concerning checksums.
Here is what I am seeing
I started with mymodel (model_mon19), ran one round of refinement against
newdatadmm2_unique1.mtz in phenix 1.4.3 with command line like the one below
.
Using the output pdb , I rebuilt a lot and renamed pdb to model-fri24.pdb ,
Now when I try refine using the same input mtz ..phenix complains about
rfree checksum mismatch again.
I am assuming that phenix does use the rfree from the mtz file
(newdatadmm2_unique1.mtz) and the checksum for the rfree entered in the pdb
should have come during round 1 from newdatadmm2_unique1.mtz , so I am a
little concerned that it complains.
The command lines I am using are given below: I only change the model , the
ncs definition file and the directory in which the refinement is run.
between runs.
Hari
phenix.refine newdatadmm2_unique1.mtz model-fri24.pdb
simulated_annealing=true simple_ncs_from_pdb.ncs refinement.main.ncs=true
refinement.input.xray_data.r_free_flags.generate=False --overwrite
ordered_solvent=true main.number_of_macro_cycles=10
strategy=rigid_body+group_adp+individual_sites
refinement.ncs.excessive_distance_limit=None
refinement.input.experimental_phases.labels=newdatadmm2_unique1.mtz:HLAC,HLBC,HLCC,HLDC>phenix_fri24.log
Command that wrote the checksum is same as above with model replaced with a
model that represented my initial build , untouched by phenix.
On Tue, Apr 21, 2009 at 8:38 PM, Ralf W. Grosse-Kunstleve
Most other things weere the same including the resolution cutoff used and the input mtz. But phenix tells me that the md5checksums dont match.
Sorry, that's unexpected. The cutoff doesn't matter. The checksum is computed on the input data, before the cutoff is applied. The checksum is just a safety guard to avoid accidents. If you are sure the test set has not changed, simply remove the remark with the checksum from the pdb file. Let me know if the problem happens again and I'll take a closer look. In that case I will need all the input files and the sequence of commands to reproduce the problem (including the command that wrote the pdb file with the checksum).
Ralf _______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb
Please disregard my previous message . The mistake was entirely mine
I had switched pdb files midstream and so the checksums got tangled up.
The checksum in the pdb came from an earlier round of refinement when I had
asked phenix to generate an rfree .
How does one get at the rfree flags that phenix uses when you ask it to
generate it. That way I can go back to these values and carry out step two
of my refinement.
Thanks for your help..and sorry for the confusion
Hari
On Fri, Apr 24, 2009 at 10:55 AM, hari jayaram
Hello Ralf and eversone else Thanks for your email concerning checksums.
Here is what I am seeing
I started with mymodel (model_mon19), ran one round of refinement against newdatadmm2_unique1.mtz in phenix 1.4.3 with command line like the one below . Using the output pdb , I rebuilt a lot and renamed pdb to model-fri24.pdb , Now when I try refine using the same input mtz ..phenix complains about rfree checksum mismatch again.
I am assuming that phenix does use the rfree from the mtz file (newdatadmm2_unique1.mtz) and the checksum for the rfree entered in the pdb should have come during round 1 from newdatadmm2_unique1.mtz , so I am a little concerned that it complains.
The command lines I am using are given below: I only change the model , the ncs definition file and the directory in which the refinement is run. between runs.
Hari
phenix.refine newdatadmm2_unique1.mtz model-fri24.pdb simulated_annealing=true simple_ncs_from_pdb.ncs refinement.main.ncs=true refinement.input.xray_data.r_free_flags.generate=False --overwrite ordered_solvent=true main.number_of_macro_cycles=10 strategy=rigid_body+group_adp+individual_sites refinement.ncs.excessive_distance_limit=None refinement.input.experimental_phases.labels=newdatadmm2_unique1.mtz:HLAC,HLBC,HLCC,HLDC>phenix_fri24.log
Command that wrote the checksum is same as above with model replaced with a model that represented my initial build , untouched by phenix.
On Tue, Apr 21, 2009 at 8:38 PM, Ralf W. Grosse-Kunstleve < [email protected]> wrote:
Most other things weere the same including the resolution cutoff used and the input mtz. But phenix tells me that the md5checksums dont match.
Sorry, that's unexpected. The cutoff doesn't matter. The checksum is computed on the input data, before the cutoff is applied. The checksum is just a safety guard to avoid accidents. If you are sure the test set has not changed, simply remove the remark with the checksum from the pdb file. Let me know if the problem happens again and I'll take a closer look. In that case I will need all the input files and the sequence of commands to reproduce the problem (including the command that wrote the pdb file with the checksum).
Ralf _______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb
Hi Hari,
How does one get at the rfree flags that phenix uses when you ask it to generate it. That way I can go back to these values and carry out step two of my refinement.
phenix.refine writes a file <prefix>_data.mtz, which contains all the original input data used in refinement and the R-free flags (original input or newly generated). Ralf
participants (2)
-
hari jayaram -
Ralf W. Grosse-Kunstleve