Re: [phenixbb] Real space difference maps?

Hi Oliver, ok, this makes sense! I will add option to output difference map and let time prove its usefulness! All the best, Pavel On 3/15/16 12:26, Oliver Clarke wrote:

Hi Pavel,

Yes, you can see it in the original map, of course, but it is still useful as a visual aid while building, and to do rigid body fitting of ligands into the difference density (where otherwise they might stray into the protein density at low resolution).

Cheers, Oli

On Tue, Mar 15, 2016 at 2:20 PM, Pavel Afonine mailto:[email protected]> wrote:

Hi Oliver,

something I'm not quite sure I understand.. if you can see the ligand in the difference map you should be able to see it in the original experimental map too (at appropriate contour level).

It's different from crystallography, I think. In crystallography we almost always use model phases to compute usual maps, meaning we have to deal with model bias and other artifacts. In cryo-EM the reconstructed map is your primary data for atomic modeling. If there is signal you should be able to see it in this map.

Pavel

On 3/15/16 12:03, Oliver Clarke wrote:

...

Hi Pavel, Thanks! Such maps can definitely be helpful, I calculate them quite often - to identify bound ligands it is certainly helpful, or to highlight regions of the model that have poor fit to the map, similar to their use in crystallography.

Cheers, Oliver.

On Tue, Mar 15, 2016 at 12:47 PM, Pavel Afonine mailto:[email protected]> wrote:

Hi Oliver,

why not.. I think we can compute map from model and then subtract it from the experimental input map (after scaling them somehow).

No I did not have that plan because it wasn't clear to me how such map can be helpful (especially since we don't do best possible job on B factor refinement).

I will add this option.

Pavel

On 3/15/16 10:25, Oliver Clarke wrote:

Hello,

is there any possibility of (or current plan for) including the capacity in a future PHENIX release to calculate a residual map automatically after phenix.real_space refine, (i.e. simulating a map from the model incorporating corrections for B-factor variation and calculating a difference map with the experimental map)? This would be very convenient when building into EM maps.

Cheers, Oli