Re: [ccp4bb] Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified?
On 25/08/10 23:24, Garib Murshudov wrote: However 1) coot uses 2mFo-DFc maps Typically yes, but it uses whatever Refmac (or other programs) provide (of course) 2) you should be able to feed any map you want to coot Yes. so it is nice place for experimenting this kind of calculation Hopefully. 3) You may try to relax gemetry R/RC - Refinement Weight. (I often change it to 6 (from 60) if I need to tighten things up.) 4) Usually if the model does not fit into electron density and programs do not want to correct it then there may be some fundamental problem in this part of the model Agreed! On 25/08/10 23:13, Hailiang Zhang wrote: Actually I tried coot real space refine zone, but the model seems not sliding into the best density map Can you expand on that? On 25/08/10 23:33, Pavel Afonine wrote: Compare slides #4 and #5: http://cci.lbl.gov/~afonine/rsr.pdf Yes, slide 5 is what Coot uses. Paul.
Re: [ccp4bb] Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified?
Well - that is exactly what COOT does isnt it? Eleanor Hailiang Zhang wrote: Hi, Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified? By the way, I have registered phenix bb. Just didn't realize this before, sorry again. Best Regards, Hailiang
[ccp4bb] Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified?
Hi, Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified? By the way, I have registered phenix bb. Just didn't realize this before, sorry again. Best Regards, Hailiang
Re: [ccp4bb] Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified?
Why you do not use coot? It does exactly what you want. regards Garib On 25 Aug 2010, at 22:33, Hailiang Zhang wrote: Hi, Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified? By the way, I have registered phenix bb. Just didn't realize this before, sorry again. Best Regards, Hailiang
Re: [ccp4bb] Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified?
Hi Garib: Actually I tried coot real space refine zone, but the model seems not sliding into the best density map (I also tried dragging it around, but still not working fine). Then I found some comments saying minimizing the difference between 2mFo-DFc and Fc may be better, thats why I am asking for this. Best Regards, Hailiang Why you do not use coot? It does exactly what you want. regards Garib On 25 Aug 2010, at 22:33, Hailiang Zhang wrote: Hi, Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified? By the way, I have registered phenix bb. Just didn't realize this before, sorry again. Best Regards, Hailiang
Re: [ccp4bb] Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified?
I mean the density of 2mFo-DFc or Fc maps. On Wednesday 25 August 2010 03:13:53 pm Hailiang Zhang wrote: Hi Garib: Actually I tried coot real space refine zone, but the model seems not sliding into the best density map (I also tried dragging it around, but still not working fine). Then I found some comments saying minimizing the difference between 2mFo-DFc and Fc may be better, thats why I am asking for this. Wait a minute. If you are minimizing a residual based on differences between Fo and Fc (never mind the precise coefficients), how is this real space refinement? (puzzled) Ethan Best Regards, Hailiang Why you do not use coot? It does exactly what you want. regards Garib On 25 Aug 2010, at 22:33, Hailiang Zhang wrote: Hi, Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified? By the way, I have registered phenix bb. Just didn't realize this before, sorry again. Best Regards, Hailiang -- Ethan A Merritt Biomolecular Structure Center, K-428 Health Sciences Bldg University of Washington, Seattle 98195-7742
Re: [ccp4bb] Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified?
On Wednesday 25 August 2010 03:19:53 pm zhan...@umbc.edu wrote: I mean the density of 2mFo-DFc or Fc maps. I still don't understand. Real space refinement will minimize a fit of model to density in whatever map you give it. It's up to you which coefficients are used to calculate the map you are refining against. Ethan On Wednesday 25 August 2010 03:13:53 pm Hailiang Zhang wrote: Hi Garib: Actually I tried coot real space refine zone, but the model seems not sliding into the best density map (I also tried dragging it around, but still not working fine). Then I found some comments saying minimizing the difference between 2mFo-DFc and Fc may be better, thats why I am asking for this. Wait a minute. If you are minimizing a residual based on differences between Fo and Fc (never mind the precise coefficients), how is this real space refinement? (puzzled) Ethan Best Regards, Hailiang Why you do not use coot? It does exactly what you want. regards Garib On 25 Aug 2010, at 22:33, Hailiang Zhang wrote: Hi, Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified? By the way, I have registered phenix bb. Just didn't realize this before, sorry again. Best Regards, Hailiang -- Ethan A Merritt Biomolecular Structure Center, K-428 Health Sciences Bldg University of Washington, Seattle 98195-7742 -- Ethan A Merritt Biomolecular Structure Center, K-428 Health Sciences Bldg University of Washington, Seattle 98195-7742
Re: [ccp4bb] Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified?
Ethan, I still don't understand. Real space refinement will minimize a fit of model to density in whatever map you give it. It's up to you which coefficients are used to calculate the map you are refining against. This is true. However, I guess, the way it is implemented in Coot is a bit different (Paul, please correct me if I'm wrong). Compare slides #4 and #5: http://cci.lbl.gov/~afonine/rsr.pdf Pavel.
Re: [ccp4bb] Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified?
I think Paul (Emsley) is better qualified to answer to this question. However 1) coot uses 2mFo-DFc maps 2) you should be able to feed any map you want to coot so it is nice place for experimenting this kind of calculation 3) You may try to relax gemetry 4) Usually if the model does not fit into electron density and programs do not want to correct it then there may be some fundamental problem in this part of the model a) multiple conformation b) out of register errors c) pepflip or something like that. If it is extremely necessary we can add such option in ccp4. Regards Garib On 25 Aug 2010, at 23:13, Hailiang Zhang wrote: Hi Garib: Actually I tried coot real space refine zone, but the model seems not sliding into the best density map (I also tried dragging it around, but still not working fine). Then I found some comments saying minimizing the difference between 2mFo-DFc and Fc may be better, thats why I am asking for this. Best Regards, Hailiang Why you do not use coot? It does exactly what you want. regards Garib On 25 Aug 2010, at 22:33, Hailiang Zhang wrote: Hi, Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified? By the way, I have registered phenix bb. Just didn't realize this before, sorry again. Best Regards, Hailiang
Re: [ccp4bb] Can some utilities of CCP4 do the real-space refinement locally with the residue range explicitly specified?
But equation given in slide #4 is exactly least-square equation with
some modified maps. Just use Pareval's theorem, then for case of 2mFo-
Dfc you will have
sum_{reflection used) (2mFo-DF_{c current) -k F_{model})^2
F_model is equal to F_{c current} at the point of calculation. All
gradients can be calculated analytically.
That is an interesting observation.
regard
Garib
On 25 Aug 2010, at 23:33, Pavel Afonine wrote:
Ethan,
I still don't understand. Real space refinement will minimize a fit
of model to density in whatever map you give it. It's up to you
which coefficients are used to calculate the map you are refining
against.
This is true.
However, I guess, the way it is implemented in Coot is a bit
different (Paul, please correct me if I'm wrong).
Compare slides #4 and #5:
http://cci.lbl.gov/~afonine/rsr.pdf
Pavel.
