Re: [phenixbb] Missing sanity check of duplicate resids within one chain?

27 Aug 2014

      Dale, Nat:

thanks for the helpful comments. I was not aware that this is a “feature” of the PDB standard.

Since Phenix recognizes the issue and issues a warning I assume this situation is handled correctly from the refinement side and does not cause any problems.

However, at least from my end, I would like to avoid this sort of issue from the get-go. The default setting in Phenix seems to be that the program continues despite the warnings. If one operates like me and does not go through the quite lengthy log files to look for these potential issues, then the likelihood of them being missed is quite high. As a suggestion: one could either think of implementing a “strict” mode in Phenix where the program aborts on encountering a warning or print out a warning at the end of the refinement that issues/warnings were encountered warranting a follow-up.

Cheers,

                Carsten

From: Nathaniel Echols [mailto:[email protected]]
Sent: Tuesday, August 26, 2014 8:44 PM
To: Schubert, Carsten [JRDUS]
Cc: [email protected]
Subject: Re: [phenixbb] Missing sanity check of duplicate resids within one chain?

On Tue, Aug 26, 2014 at 2:51 PM, Schubert, Carsten [JRDUS] mailto:[email protected]> wrote:

Phenix (1.9-1692) seemingly happily accepted the input and proceeded with refinement and validation. After upgrading my coot installation to the latest build the error was flagged.  I think this sort of mistake should have at least been flagged in the validation stage or is this out of scope for phenix?

As Dale notes, this is technically allowed by the PDB format - however, we do already look out for it:

nat@laptop> phenix.pdb.hierarchy carsten.pdb
. . .
  ### WARNING: consecutive residue_groups with same resid ###
  number of consecutive residue groups with same resid: 1
    residue group:
      "HETATM 9327 CL    CL S   3 .*.    Cl1-"
    next residue group:
      "HETATM 9328  K     K S   3 .*.     K1+"

We could certainly present this warning elsewhere - the question is where would it be most likely to be noticed?

-Nat