Hi Simome,

I get a very nice difference density for the ligand:

http://cci.lbl.gov/~afonine/tmp/coot.png

Command I used:

phenix.real_space_diff_map 5umd_box.pdb 5umd_box.ccp4 resolution=3.2

Make sure to use map_model_difference_1.ccp4 as it seems to be better than the seconds map and also load it into Coot as a difference map.

So.. all seems to work fine for me.

Pavel

On 3/14/17 08:42, Simone Pellegrino wrote:

Dear Pavel,

I am trying to obtain a difference map for the recent pdb ID: 5umd. I took the map deposited as emd_8576 and tried to obtain the difference map as you suggest. Unfortunately I have several blobs appearing but nothing where the ligand (an antibiotic binding the ribosome) was (I removed it from the initial pdb prior running the script). Also I find that it is difficult to interpret the two maps coming as output.

Many thanks for your comments on this.

Best

Simone

2016-12-03 4:54 GMT+01:00 Pavel Afonine <pafonine@lbl.gov>:

For those practicing cryo-EM and tending to use Phenix nightly builds, as well as in response to a recent post (Re: [phenixbb] difference map based on the cryo-EM map):

Next Phenix nightly build (dev-2612 and up) will have a command line tool to compute map-model real-space difference maps.

One map is computed in real space: cryo-EM map minus model calculated map; it is locally scaled in some complex way to account for local variations and possibly poorly refined B-factors. In terms of reciprocal space it is an analog of:

(F_obs, Phase_obs) - (F_calc, Phase_calc).

The other map is computed using both spaces, and is analog of (F_obs-F_calc, Phase_obs); here no model phase is involved.

I'm not sure which one of the two is better.

Both maps are useful to locate ligands or yet not built pieces of the model (very much like in crystallography).

Current limitations:
- no GUI;
- map must have origin at (0,0,0);
- output file name can't be changed.

To use:

phenix.real_space_diff_map model.pdb map.ccp4 resolution=3.9

which will output two maps: map_model_difference_1.ccp4 and map_model_difference_2.ccp4 .

If proving ligand is the purpose to compute this map, then obviously it should not be present in model.pdb.

In future (once fully tested) this will be integrated into phenix.real_space_refine.

The tool is 10 hours old and tested on just one model (5l4g, emd_4002.map), so bugs are not unexpected. Any feedback is welcome!

All the best,
Pavel

_______________________________________________
phenixbb mailing list
phenixbb@phenix-online.org
http://phenix-online.org/mailman/listinfo/phenixbb
Unsubscribe: phenixbb-leave@phenix-online.org

--

Dr. Simone Pellegrino
I.G.B.M.C.
1, Rue Laurent Fries
67404 Illkirch Cedex
France