Hi Maxime,

this problem is now addressed in phenix.refine and other related tools (such as phenix.pdbtools, phenix.model_vs_data, phenix.geometry_minimization): next available nightly build should have it. The correct distances are used based on "scattering_table=neutron" flag in phenix.refine and phenix.model_vs_data, and "use_neutron_distances=true" in phenix.pdbtools and phenix.geometry_minimization.

Please let me know should you have any questions and notice any problems.

Thanks,
Pavel

On 11/6/12 9:45 AM, Maxime Cuypers wrote:

Hi,

i am refining a perdeuterated neutron structure and have noticed a lot of positive density aside the D atoms on aliphatic and aromatic side chains. just in the alignment of C-D bonds (C-D then +), suggesting that phenix.refine constraints these bond lengths too short in my case. i tried automatic weights for geometry refinement and unrestrained specific C-D refinement but it is not satisfactory.

then i found the following funny distances (angströms) in the pdb database neutron structures:

3KYX:
program: Phenix.refine
D-prot
Distances:
C-D aromatics = 0.93-0.94
C-D aliph. = 0.98

1CQ2
Program : xplor3.1
D-protein (despite the atoms being labelled as H according to abstract)
Distances:
C-D aromatics = 1.06-1.08
C-D aliph. = 1.08

4FC1
H-protein
Distances:
program: Phenix.refine
C-H aromatics = 0.93
C-H aliph. = 0.97

1VCX
H-protein
program: CNS
Distances:
C-H aromatics = 1.06-1.10
C-H aliph. = 1.09

those neutron C-D/C-H bond lengths in pdb seem pretty much inconsistentto me.

So it seems that phenix.refine uses C-D/ C-H bonds restrains that are different to other programs to below 1.0 Ang. ( ca. 0.10 A differences)
how can i change the restraints to make it all longer bond lengths for C-D only? with a .cif file?
i saw that phenix uses the CCP4 monomer library but the one in :
/usr/local/phenix-.../chem_data/mon_lib/list
only seem to be applying modifications.

any clue on how to specifically modify the restraints parameters on C-D bonds would be very appreciated.

Thanks,

Maxime
_______________________________________________
phenixbb mailing list
[email protected]
http://phenix-online.org/mailman/listinfo/phenixbb