So the balance here will be with what Jmol will do and what people want
to customize it to do (at web pages, for example).

that sounds OK to me (I'm mostly interested in web pages)
 


Mauricio Carrillo Tripp wrote:

> This is great Bob, and is exactly what I was after last month
> (http://www.mail-archive.com/jmol-users@lists.sourceforge.net/msg10041.html),
> my apologies if I was not clear enough then ;)
> Now, as for an spherical capsids 'user' at VIPERdb, I would like to be
> able to:
>
> 1) have the option to show/hide any of the biomolecules in the PDB
> file, in spherical capsids there's always 60. This doesn't mean all 60
> biomolecules should be loaded initially by default, but only when the
> user chooses to. The default will always be to load and show
> biomolecule number one. As I understand it, every time the user wants
> to display another biomolecule in the biounit, the PDB file needs to
> be (down)loaded again and read using a different filter? If so, I
> think it would be more practical (but perhaps more trouble to
> implement) to load the PDB file once, save all the biomats
> corresponding to all the biomolecules in the biounit, and then use
> them latter as needed? (this might save a lot of HTTP requests ;) I'm
> thinking about buttons to show 1/4, 1/2 or full capsid. But also an
> option to show biomols 1,6 and 7 (or any combination), to present an
> specific interface.
>
By default, Jmol will simply load the coordinates -- that's not
negotiable. We could imagine a script "set" flag, though, that would
direct Jmol to some other behavior.

The PDB file does not need to be downloaded again in terms of a browser
-- the browser will cache the file, and it will load very fast from the
hard drive. Right now the biomolecules (each including biomt matrix
array and chain list) are being saved, and in principle they could be
accessed later. We can look into this option. In the mean time you will
need to do a re-load. But we have load APPEND, and it is very fast with
reasonably sized systems, so that might work better than you think.

OK
 

Buttons to show "1/4, 1/2, or full capsid" would have to be set up by a
user (you). There are ways to customize the Jmol popup menu -- with a
simple small file on disk that expands a menu item. Or if using a web
page, it would be an HTML button, perhaps.

yes, that's what I meant, building (us) an ad hoc UI for the Jmol applet,
all we need is the functionality and the API ;)
 
The different sets of atoms
created by individual BIOMT transformations are selectable using the
"symop" selection option:

select symop=3
select symop<6

etc. (There is a bug there -- symop=1 specifically is not selectable in
11.5.32)

OK, so for example, if I want to display 1/2 capsid, I will have to load
the PDB file 30 times, each time using a different select in the loading filter.
By using the append option in the load command, all 30 biomolecules
will 'live' on different frames, which can be independently selected by
frame, OR the symop operator. correct?
 


> 2) have the option to apply symmetry to any number of chains in the
> ASU (many capsids have more than one chain in the ASU), I believe one
> can do that with the filter as is, right? I'm thinking of the option
> to show chain A of biomol 1 and chain C of biomol 7 (or any
> combination), again, to present an specific interface.
>
So tell me more about what it means when there are multiple
biomolecules. What would that indicate?

The spherical virus capsid (biounit) is made up of 60 identical copies of the ASU (biomol 1).
The ASU in itself can be made up of one or more independent chains, which in general
terms corresponds to the triangulation number value (T). So for example, a T=3 capsid has
3 independent chains (A,B,C) forming the ASU, which in turn will give 180 (3x60) chains forming the
full capsid. The PDB file contains coordinates for the first copy of the ASU, and also the 60
transformations (biomat 1...60) that need to be applied to those coordinates in order to generate the full
capsid (usually the first transformation is the identity). So each PDB file, in essence, contains a full
capsid (functional biounit), in a very reduced space (file size).
In addition to the intra-unit interfaces (protein-protein interactions between the chains that form the
ASU), one is also interested in seeing/studying the inter-unit interfaces (protein-protein interactions
between the chains of different copies of the ASU). Some of these inter-unit interfaces are formed by
up to six copies of the ASU, all sharing a common symmetry axis (in this case, a 6-fold symmetry axis).
 
The way you would do this
currently would be to load biomolecule 1 into one Jmol frame and
biomolecule 7 into another frame. Sounds like you would like to be able
to display these independently or together - not a problem. I'm thinking
this would be a menu option to load or to load APPEND.

> 3) have the option to apply symmetry just to a segment of an specific
> chain in the ASU, I believe one can do that with the filter as is,
> right? I'm thinking about the option to show only the residues
> involved in an specific quaternary structure, beta annulus for
> example, around an icosahedral symmetry axis, comprised from a few
> residues of chain B of biomol 1, a few from chain C of biomol 6, and
> so on ...
>
Is your interest the applet? This sounds like what Jmol.js is for.

yes, I agree.
 


> I'm thinking on implementing an ad hoc UI for the above, so I don't
> know about the Jmol menu. Although I guess for other systems it would
> be nice to have a list of all biomols available.
>
> 4) have all the biomolecules (as whole or parts, as explained above)
> as different model/frame to apply different representations to them.
> So, "load append" works?
>
oh, yes. load append is great. Well tested. The big issue is going to be
memory.

In most cases there is no need to generate a full capsid (it looks pretty awesome
though ;). Because of symmetry, ~1/4 capsid contains all interfaces of interest.
 


> Some, or all, of the points above might be valid for other systems
> too. As you said Bob, this is huge, and I'm very exited about it, will
> certainly simplify the scripts we use, and more importantly, the way
> people interact with these structures and learn from them.
>
> Thanks again for the great work!
>
my pleasure!


--
0 | Mauricio Carrillo Tripp, PhD
/ | Department of Molecular Biology, TPC6
0 | The Scripps Research Institute
\ | 10550 North Torrey Pines Road
0 | La Jolla, California 92037
/ | trippm@scripps.edu
0 | http://viperdb.scripps.edu/