[phenixbb] Fitting ligand using coot

[Nathaniel Echols]

On Sun, Mar 31, 2013 at 10:09 AM, Bishwa Subedi <bsubedi at btk.fi> wrote:
> The bigger problem is with the refinement even after fitting the ligand. As
> suggested I have figured out with fitting the ligand. I performed real space
> refinement to the ligand and it fits well. I used the merge function in coot
> to merge ligands with molecules (I have two molecules in asymmetric unit ).
> Now the problem is with the phenix.refine. The CIF file is not accepted.

I don't know what "not accepted" means - what happened when you tried
to load it?  The files you sent me loaded and ran without any problems
in both version 1.8.2 and the current code.

The problem with the structure you fed to ReadySet is probably that
the residue name is changed to LIG, which is a generic residue code
that we strongly discourage using.  Because you didn't supply any
other CIFs and LIG doesn't uniquely identify the chemical component,
ReadySet has to guess the molecular topology, which is very difficult
to do from heavy-atom coordinates alone.  I'm not at all surprised
that this crashes.  (It also screws up the chirality, but if you use
the original CIF and residue name this shouldn't happen.)

A general plea to the list (repeating what Pavel has said many times):
when reporting problems with any part of Phenix, send us the *exact*
error message you receive, along with any Python traceback (this may
not always be shown).  Generic descriptions like "it crashed" or "it
didn't work" tell us nothing and lead to endless email exchanges.  I
know from experience that many users worry about filling up our
inboxes with irrelevant details, but for the record: the more
information you send, the better, and we would much rather hear when
our code doesn't work properly than have you give up in frustration
(or switch to another program).  The only thing you should avoid is
sending massive attachments to the entire list (although the server
will block these anyway).

-Nat