Hi Carsten, they should definitely add up to 1. If the PDB file contains altloc identifiers, then those atoms should be picked up automatically and you don't need to specify any selections. For example, make sure you have something like this in your PDB file (the starting occupancies can be any): ATOM 49 N MET A 10 3.832 -11.001 0.570 1.00 11.55 N ATOM 50 CA AMET A 10 2.523 -10.377 0.709 0.70 12.44 C ATOM 51 CA BMET A 10 2.574 -10.279 0.718 0.30 12.42 C ATOM 54 CB AMET A 10 2.580 -9.129 1.601 0.70 14.67 C ATOM 55 CB BMET A 10 2.822 -8.921 1.394 0.30 13.99 C ATOM 60 CG AMET A 10 1.245 -8.462 1.951 0.70 18.00 C ATOM 61 CG BMET A 10 1.545 -8.166 1.727 0.30 17.90 C ATOM 66 SD AMET A 10 0.659 -7.204 0.780 0.70 19.47 S ATOM 67 SD BMET A 10 1.792 -6.615 2.600 0.30 26.45 S ATOM 68 CE AMET A 10 1.520 -5.730 1.337 0.70 20.21 C ATOM 69 CE BMET A 10 2.371 -5.566 1.271 0.30 26.12 C ATOM 76 C MET A 10 2.021 -10.028 -0.671 1.00 10.57 C ATOM 77 O MET A 10 2.780 -9.724 -1.578 1.00 11.30 O and then everything should be OK. Unfortunately, I can't tell much about the selections you use without looking at the PDB file, so in your case I would just try the above suggestion. Pavel. On 2/4/2009 12:15 PM, Schubert, Carsten [PRDUS] wrote:
Hi,
I am trying to sort out some issues with the constrained occupancy refinement in phenix 1.4.3. The structure I am working on is fairly high resolution ~1.4A and it looks as if a couple of the alternative conformations would benefit from occupancy refinement, no hydrogens are present yet.
I defined the constrained groups like this:
occupancies { individual = None remove_selection = None constrained_group { selection = chain A and resid 66 and altid A selection = chain A and resid 66 and altid B } ....
I expected that the sum of occupancies would be restrained to 1, which they should according to the manual, but the refinement gives me occupancies of 0.7 for altid A and 0.34 for altid B, starting values were both 0.5. Similar discrepancies were also present in other groups, like 0.5 A, 0.37 B etc. The latter is the largest deviation from 1 I have seen in thus far. The alternate conformations were generated in coot, by splitting at the CA position.
Is this a rounding error I just have to contend with or is there some other parameter I need to tweak?
Cheers,
Carsten
_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb