[phenixbb] Very poor density, but very important loops

Pavel Afonine PAfonine at lbl.gov
Fri Oct 23 13:14:48 PDT 2009

Hi Andrew,

we (Lester and myself) recently implemented a new tool in PHENIX called 
It is based on the idea published a while ago:

Acta Cryst. (1997). D53, 540-543  
Local Improvement of Electron-Density Maps
A. G. Urzhumtsev

We are currently looking for tests cases for this procedure, and I guess 
your case is just perfect for it!

The phenix.grow_density tool is not at that level of automation so you 
can take it, run it and get the answer, but rather it would require some 
back-and-forth trials.

So, if you would like to give it a try, here is what I suggest:

- you send me the data and model and tell what are the loops you would 
like to improve (of course all data will be kept confidentially);
- I run it through phenix.grow_density and hopefully (no promises, we 
are testing it!) give you back improved maps that you can use to build 
in your poor loops.
- We may need to go through a few iterations of this (or may not if it 
doesn't work from the start).

I will be in travel next week, so if you decide to send the data, I 
can't promise that I work on it immediately, but I (or Lester) will try.

Please let me know what you think.

Of course, you can try computing kick maps as well to see if it improves 
the maps too.


On 10/23/09 11:57 AM, Waight, Andrew wrote:
> Hello All, 
>                I am putting the finishing touches on my structure, a novel membrane protein channel. The data extends to 2.1A and the R factors are currently at ~17/21. However there is a certain loop/helical region where the density gets pretty poor, Unfortunately this is a relatively important ligand binding site so I would like to extract as much information from my patchy blobs as possible. I am wondering if generating a kick map would be useful in this scenario. If so am I missing something other than setting use_kick_maps = True and adding a couple more map definition labels (as described by Pavel)? Also does anyone have any advanced technique suggestions for this type of circumstance, i.e. would it make sense to combine multiple reflection datasets? Thanks so much for all your help, I am a true phenix enthusiast and feel like I owe quite a bit to all of you who have made this amazing program possible. 
>                                               Drew Waight

More information about the phenixbb mailing list