Dear phenix.refine community: I am solving a protein structure using experimental phasing. In order to do that I combined two derivative isomorphous datasets into a double SAD experiment (to increase the individual weak anomalous power from each dataset) using SHARP.My resolution is 3.2 A, and the space group is C2221. I managed to build a model using the density-modified phases from SHARP. Then I started refining against one of my derivative datasets because my native dataset is non-isomorphous due to the heavy meal soaking.Then I made two mistakes :( 1. I did not realize at the beginning of my refinement-building rounds that I was refining against the dataset that showed the worst statistics. 2. I was using a dataset with only FP and SIGFP but not DANO and SIGDANO. I used the phases from my SHARP map (also the experimental phase restraints) Now I did a new run of refinement using my best dataset (better statistics) and also including the anomalous data FP, SIGFP, DANO and SIGDANO. Obviously, my R-work and R-free got pretty much better. Nevertheless, the R-free was better than R-work at the beginning of the run, at the end of the refinement R-free and R-work inverted as usually (being R-free worst than R-work) r_work = 0.2582 r_free = 0.2888 . Certainly, there was a mistake with a new assignment of the free set of reflections. Now I am aware that the set of free reflections has to be the same all over the refining procedure. Here are the questions How can I be sure that my R-free is still a safe cross validation of the correctness of my structure? Is it useful to perform a round of simulated annealing in order to erase any wrong information in my model and start the refinement rounds again? Do you have any suggestions? All the best Jose Farias
On Thu, Sep 1, 2011 at 9:25 AM, Jose Arcadio
Now I did a new run of refinement using my best dataset (better statistics) and also including the anomalous data FP, SIGFP, DANO and SIGDANO. Obviously, my R-work and R-free got pretty much better.
Nevertheless, the R-free was better than R-work at the beginning of the run, at the end of the refinement R-free and R-work inverted as usually (being R-free worst than R-work) r_work = 0.2582 r_free = 0.2888 . Certainly, there was a mistake with a new assignment of the free set of reflections. Now I am aware that the set of free reflections has to be the same all over the refining procedure.
Here are the questions
How can I be sure that my R-free is still a safe cross validation of the correctness of my structure?
The behavior of R-work and R-free suggest to me that the second dataset is isomorphous with the first - is this correct? If so, you can transfer the R-free flags between files using the reflection file editor in the GUI. It will automatically extend the original flags to higher resolution if necessary. I think there is also a way to do this within phenix.refine, but I haven't tested it.
Is it useful to perform a round of simulated annealing in order to erase any wrong information in my model and start the refinement rounds again?
Yes, this is the next best choice if you can't preserve the original test set. But you should try to copy it over first. -Nat
Hi Jose, DANO and SIGDANO are not used in refinement. Resetting ADPs to an average value, and performing ~5 macro-cycles of refinement with SA and weights optimization typically should be sufficient. Make sure you are using the latest PHENIX version from nightly builds: http://www.phenix-online.org/download/nightly_builds.cgi Let me know if you have any questions. Pavel. On 9/1/11 9:25 AM, Jose Arcadio wrote:
Dear phenix.refine community:
I am solving a protein structure using experimental phasing. In order to do that I combined two derivative isomorphous datasets into a double SAD experiment (to increase the individual weak anomalous power from each dataset) using SHARP.My resolution is 3.2 A, and the space group is C2221.
I managed to build a model using the density-modified phases from SHARP. Then I started refining against one of my derivative datasets because my native dataset is non-isomorphous due to the heavy meal soaking.Then I made two mistakes :(
1. I did not realize at the beginning of my refinement-building rounds that I was refining against the dataset that showed the worst statistics.
2. I was using a dataset with only FP and SIGFP but not DANO and SIGDANO. I used the phases from my SHARP map (also the experimental phase restraints)
Now I did a new run of refinement using my best dataset (better statistics) and also including the anomalous data FP, SIGFP, DANO and SIGDANO. Obviously, my R-work and R-free got pretty much better.
Nevertheless, the R-free was better than R-work at the beginning of the run, at the end of the refinement R-free and R-work inverted as usually (being R-free worst than R-work) r_work = 0.2582 r_free = 0.2888 . Certainly, there was a mistake with a new assignment of the free set of reflections. Now I am aware that the set of free reflections has to be the same all over the refining procedure.
Here are the questions
How can I be sure that my R-free is still a safe cross validation of the correctness of my structure?
Is it useful to perform a round of simulated annealing in order to erase any wrong information in my model and start the refinement rounds again?
Do you have any suggestions?
All the best
Jose Farias
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
On Thu, Sep 1, 2011 at 9:49 AM, Pavel Afonine
DANO and SIGDANO are not used in refinement.
Not entirely true - if the input data are FP,SIGFP,DANO,SIGDANO, the original F+ and F- will be reconstructed and used for refinement. (If you're not refining anomalous groups, however, this is unlikely to make much of a difference.) -Nat
On 9/1/11 6:49 PM, Pavel Afonine wrote:
Hi Jose,
DANO and SIGDANO are not used in refinement.
Resetting ADPs to an average value, and performing ~5 macro-cycles of refinement with SA and weights optimization typically should be sufficient. Make sure you are using the latest PHENIX version from nightly builds: http://www.phenix-online.org/download/nightly_builds.cgi
Let me know if you have any questions.
Pavel.
On 9/1/11 9:25 AM, Jose Arcadio wrote:
Dear phenix.refine community:
I am solving a protein structure using experimental phasing. In order to do that I combined two derivative isomorphous datasets into a double SAD experiment (to increase the individual weak anomalous power from each dataset) using SHARP.My resolution is 3.2 A, and the space group is C2221.
I managed to build a model using the density-modified phases from SHARP. Then I started refining against one of my derivative datasets because my native dataset is non-isomorphous due to the heavy meal soaking.Then I made two mistakes :(
1. I did not realize at the beginning of my refinement-building rounds that I was refining against the dataset that showed the worst statistics.
2. I was using a dataset with only FP and SIGFP but not DANO and SIGDANO. I used the phases from my SHARP map (also the experimental phase restraints)
Now I did a new run of refinement using my best dataset (better statistics) and also including the anomalous data FP, SIGFP, DANO and SIGDANO. Obviously, my R-work and R-free got pretty much better.
Nevertheless, the R-free was better than R-work at the beginning of the run, at the end of the refinement R-free and R-work inverted as usually (being R-free worst than R-work) r_work = 0.2582 r_free = 0.2888 . Certainly, there was a mistake with a new assignment of the free set of reflections. Now I am aware that the set of free reflections has to be the same all over the refining procedure.
Here are the questions
How can I be sure that my R-free is still a safe cross validation of the correctness of my structure?
Is it useful to perform a round of simulated annealing in order to erase any wrong information in my model and start the refinement rounds again?
Do you have any suggestions?
All the best
Jose Farias
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
Thanks a for the help Pavel and Nat. In order to reset ADPs to an average value, shall I use: set_b_iso=30 ? Could you suggest a good starting point for wxc_scale and wxu_scale for my resolution (3.2 A). Or maybe I should estimate these values empirically? I would also like to include the heavy metals in my structure, in this case I guess it is important to have DANO and SIGDANO as you also suggested. Best regards Jose
Hi Jose,
In order to reset ADPs to an average value, shall I use: set_b_iso=30 ?
phenix.pdbtools model.pdb set_b_iso=30 will do it. If you have atoms with ANISOU then convert them to isotropic first: phenix.pdbtools model.pdb convert_to_isotropic.
Could you suggest a good starting point for wxc_scale and wxu_scale for my resolution (3.2 A). Or maybe I should estimate these values empirically?
No need to do this. Either you run with the default values that are determined automatically, or let the program optimize them in which case it will try an array of values and pick the good one. See relevant (and most recent) paper discussing this here: http://www.phenix-online.org/newsletter/CCN_2011_07.pdf
I would also like to include the heavy metals in my structure, in this case I guess it is important to have DANO and SIGDANO as you also suggested.
I am not sure if "reconstructed" Fobs(+) and Fobs(-) from FP and DANO is exactly the same thing as genuine original Fobs(+) and Fobs(-). In any case, all refinement does it refines against the values you give it. In case you have Fobs(+) and Fobs(-) (or FP and DANO) phenix.refine will compute anomalous difference map for you. Whether refining against Fobs(+) and Fobs(-) or F=(Fobs(+) + Fobs(-))/2 has benefits is not clear, unless you refine f' and f'' for anomalous scatterers. Pavel
On Fri, Sep 2, 2011 at 6:29 AM, Pavel Afonine
I am not sure if "reconstructed" Fobs(+) and Fobs(-) from FP and DANO is exactly the same thing as genuine original Fobs(+) and Fobs(-).
It's probably close enough for generating an anomalous difference map - apparently Phaser does not work as well with the reconstructed amplitudes, but fortunately this structure is already solved. -Nat
participants (3)
-
Jose Arcadio -
Nathaniel Echols -
Pavel Afonine