Hi Mike,

  The Chimera unit cell dialog is behaving as expected in the cases where the asymmetric unit contains half of a benzene ring.  First, parts of the asymmetric unit are not moved relative to other parts of the asymmetric unit, so as you observe the half benzenes in asymmetric unit copies sometimes lie outside the unit cell.  All unit cell dialog
just puts the center of the asymmetric unit in the unit cell box.  A second problem is that a half benzene ring from one asymmetric unit will not be joined to a half benzene ring from another asymmetric unit.  They will line up perfectly but Chimera does not see bonds between the two halfs so for instance, selecting aromatic rings won't pick those up.

  I guess the first problem could be handled by moving each connected group of atoms so its center lies in the unit cell.  I think it could lead to undesired results.  For example a protein with a bound ligand might have the ligand outside the unit cell causing it to be moved to some other asymmetric unit's binding site.  The trouble with that is that the protein and ligand are part of one molecule model in Chimera, so now you have taken the ligand in that molecule and placed it with the protein of another copy of the molecule.  That may not be what you want.

  Another answer to the first problem, is that I added to the unit cell dialog the ability to make multiple asymmetric unit copies to fill multiple unit cells.  If you make several unit cells, then the half benzenes lying outside one unit cell do join up with the half benzenes from a different unit cell.

  If you were trying to make a figure and you didn't want half-benzenes floating around -- you could simply select them and delete them (Actions / Atoms / Delete).

  The second problem is harder.  Chimera could guess that the half-benzenes should be joined since they have exactly coincident atoms.  (This relies on the strange inconsistency that the half-benzenes in the PDB files you sent contain 4 atoms instead of the expected 3 atoms.  So it doesn't actually represent an asymmetric unit since there is an extra atom that will be duplicated in other asymmetric units.)  But Chimera can't join them because covalent bonds can't be formed between different molecule objects -- it would require making a new molecule to contain all the atoms.  Creating extra molecule models to join the half benzenes is possible but weird.

    Tom

-------- Original Message --------
Subject: Re: [Chimera-users] Centro-symmetric space groups
From: Michael Day
To: Tom Goddard
Date: 9/4/09 12:41 PM
Hi Tom,

Thanks. It looks like the issue is mostly resolved. 

The is a problem with molecular units sitting at symmetry element in non-primitive cells, for now I've found it in C2/c (iat03.pdb) and Fdd2 (ta21.pdb) which I've attached. As far as I can tell this is not unique to chimera, some programs handle this okay, many don't. You'll see that in these two cases benzene molecules land outside of the cell or fail to complete a pair.

If I had to lobby for any particular ticket on your list it would be #201. What I would add in way of description on that ticket is that I'm not talking about using SHELX in Chimera. What I meant to say is for Chimera to read a standard format CIF file containing the reciprocal space transform of the real space electron density. The fft routines to transform back to real space have been around forever and are in most all modern languages. 

All crystallography packages can write this type of file containing h,k,l F(obs)^2^, sig(F(obs)^2^), F(calc) and phi (phase angle in degrees). This the preferred format for sending data to the PDB.

I'm not looking to add a bunch of "stuff" to your already huge pile but I thought most of this was already in the public domain. After all, nothing is impossible for the guy that doesn't have to do it. It's all easy and straightforward, right? ;0)

As for the issue of reading the CIF, P21/n is a very common alternate setting of P21/c and the PDB symmetry list does not contain it.

Thanks again!!
Cheers,
Mike

<<< ------------------------------------------------------------------------>>>
Dr. Michael W. Day
Director - X-ray Crystallography Lab & Molecular Observatory
California Institute of Technology
Mail Code 139-74
Pasadena, CA 91125

<<< ------------------------------------------------------------------------>>>

On Sep 2, 2009, at 5:54 PM, Tom Goddard wrote:

Hi Mike,

  I've fixed the unit cell dialog to handle inverting symmetries.  It was a simple change.  Will be in tonight's daily builds.

  We could have Unit Cell work with CIF files but that is an enhancement that will have to compete for time with the hundreds of other feature requests we're looking at.  My list of requests is here

    http://socrates2.cgl.ucsf.edu/trac/chimera/wiki/requests

and Eric and other developer maintain their own off-line lists.  So you need to make a good case.  It could be an easy task if our third-party CIF file reader already parses the needed symmetry operators.  If it does not provide those then it could be a significant bit of work.

    Tom


-------- Original Message --------
Subject: Re: [Chimera-users] Centro-symmetric space groups
From: Michael Day
To: Thomas Goddard
Date: 9/1/09 1:19 PM
Hi Tom,

Thanks.

Another option you might consider is to have the Unit Cell tool work also with CIF files then only coordinate transformations are needed as the symmetry operators are included. An advantage is that if someone uses a non-standard setting of a space group the symmetry operators are still valid for the coordinates.

Thanks for all your work. All of you on the Chimera team (Elaine, Eric and Greg) set the bar pretty high! My hats off to you.

Cheers,
Mike

<<< ------------------------------------------------------------------------>>>
Dr. Michael W. Day
Director - X-ray Crystallography Lab & Molecular Observatory
California Institute of Technology
Mail Code 139-74
Pasadena, CA 91125

<<< ------------------------------------------------------------------------>>>

On Sep 1, 2009, at 9:27 AM, Thomas Goddard wrote:

Hi Mike,

 Thanks for the suggestions.  The coordinate transformations are routine.  The trouble is that the Chimera Molecule C++ code takes the transformation matrix and orthonormalizes it to a proper rotation.  So the inversion operator of "P -1" gets changed to inverting x and y axes but not the z axis.  The orthonormalization is supposed to just fix tiny round-off transformation problems, like matrices read from a PDB file that have only 5 digits of precision.  Really we should be warning if a transformation is very far from a proper rotation.

 I'll let you know when the problem is fixed in the daily builds.

Tom


-------- Original Message --------
Subject: Re: [Chimera-users] Centro-symmetric space groups
From: Michael Day
To: Thomas Goddard
Date: 8/31/09 7:54 PM

Tom,
Thanks again. Awesome!
I suspect mirrors and therefore glide planes could be an issue as you don't see them in proteins.
I don't know what your overall strategy is for applying space group operators but to avoid issues I would suggest;
1) deorthogonalize into crystallographic coordinates as fractions of cell edge
2) apply symmetry operators in turn to the original set (x,y,z) to get (x',y',z') for each operator
3) calculate the centroid for each equivalent set of coordinates (x',y',z')
4) move centroids into unit cell as needed so they are x=0,1 y=0,1 and z=0,1
I suspect you already know this but it never hurts to say it. My students think I overstate the basics at times but it sure helps all of us get on the same page.
Cheers,
Mike
<<< ------------------------------------------------------------------------>>>
Dr. Michael W. Day
Director - X-ray Crystallography Lab & Molecular Observatory
California Institute of Technology
Mail Code 139-74
Pasadena, CA 91125
>
<<< ------------------------------------------------------------------------>>>
On 31 Aug, 2009, at 6:14 PM, Thomas Goddard wrote:
Hi Mike,

Oops. The symmetries that invert coordinates don't work in the Unit Cell tool.  This is because it sets the transformation matrix (rotation and shift) for the asymmetric unit copies but it can only handle proper rotations.  Chimera is generally used for proteins which are handed (not symmetric under inversion) so such space groups generally don't occur for proteins.  But of course for small molecules they can.

I'll fix this soon, maybe later this week.  I'll have to change atom coordinates instead of setting the transform matrix for the asymmetric unit copies.

Tom