Herman, Yes you are right. I was assuming that since these are REMARKs then programs would ignore them and use the space group in the CRYST1 record to define the symmetry operators (a better way in my view, assuming that the space group setting symbol is unambiguous as it should be). If programs are using the SYMTRY info then of course that would also need to be updated. This would be the same transformation as for the other matrices, i.e. Ct = CoT^-1 where Co is the original SYMTRY matrix and Ct is the new one for the transformed co-ords (again 4x4 matrices).
Cheers -- Ian On 13 March 2013 15:45, <[email protected]> wrote: > ** > Dear Ian, > > My feeling is that most crystallograpic programs use indeed a separate > library with symmetry operators acting on fractional coordinates. However, > the SMTRY records in the PDB files work on the orthogonal coordinates and > some of the more advanced non-crystallographic viewing/modeling programs > use these to generate symmetry mates. In this way, those > non-crystallographic programs do not need to provide libraries with > symmetry operators and they do not need to bother whether one has a > hexagonal or rombohedral settings or whether the space group is P21212 or > P22121. If crystallographers would be using these records as well, there > would have been a lot less confusion, but that is another story. > > So to be on the safe side, one would need to either delete or transform > the SMTRY records. > > Cheers, > Herman > > > > ------------------------------ > *From:* Ian Tickle [mailto:[email protected]] > *Sent:* Wednesday, March 13, 2013 4:22 PM > *To:* Schreuder, Herman R&D/DE > *Cc:* [email protected] > > *Subject:* Re: [ccp4bb] Best practice for transformed PDB coordinates? > > > > > On 13 March 2013 14:51, <[email protected]> wrote: > >> ** >> >> As I see it, there is no need to change the orthogonal coordinates. They >> are used in connection with the SYMTRY records to generate the symmetry >> mates. Changing the orthogonal coordinates would mean changing the SYMTRY >> records, which would only complicate things. If we do not change the >> orthogonal coordinates, we do not need to change the SCALE matrix and the >> only thing we need to change is the ORIGX matrix. >> >> PS note there is no need to change the symmetry operators in either scheme, > because these actually operate on the fractional co-ordinates, not > directly on the orthogonal ones. In my scheme I redefine the SCALE > matrix so that together with the transformed co-ordinates you recover the > original fractional co-ordinates and so the symmetry still works > correctly(theoretically of course!). > > Cheers > > -- Ian > >
