Hi Nat,

It is the PHENIX.refine which I was using, but now I think the problem is due to the fact that there is an exact "copy" of atoms with the exact same coordinate but different atom number, which caused the failure for phenix.refine. I can delete those part to get the refinement going, but I still don't understand why the phenix.apply_ncs generate an extra copy.......

Thanks,

Bingfa

2014-05-07 12:42 GMT-07:00 Nathaniel Echols <nechols@lbl.gov>:

On Wed, May 7, 2014 at 12:34 PM, Sun Bingfa <sunbingfa@gmail.com> wrote:

I've been working on a structure using NCS. When I run the phenix.apply_ncs to generate the whole structure using the original one chain PDB, it gave me a PDB with same atom number on the other chain.

For example, I start with chain A and end in chain A and B (B is generated by NCS), while the atom number on chain B is just a duplicate of chain A, starting from 1. This cause error in the subsequent refinement trials.

Refinement with what program? This is certainly not a problem in phenix.refine (at least using current code); I just tried it with one of our tutorial structures.

I'm wondering what program can correct this issue.

You can run phenix.pdbtools (also in GUI, of course) on the model with no additional options, and the output will be renumbered. We can easily modify phenix.apply_ncs to do this automatically, of course, but programs really shouldn't crash on the current output.

-Nat