Re: [h5md-user] Tuples and topology, the return

[Konrad Hinsen]

Pierre de Buyl writes:

 > > But H5MD can't store what the positions mean. I don't think any PDB
 > > client would be happy to have just a bunch of positions. At the very
 > > least, you need the chemical element and the occupancy, then you
 > > have at least a useable crystallographic model.
 > 
 > Just for clarity, by "chemical element" you simply refer to its atomic
 > name/number? This is really easy to put in a module (make 'species' mandatory
 > and clarify the meaning of the species integer value).

Yes, it's easy, but it remains to be done.

 > Occupancy? If that is a single additional variable, this is also a reasonable
 > addition to H5MD (be it via a module).

It's a per-particle variable, with float values between 0 and 1.

 > > Biologists still wouldn't care.
 > 
 > Will they ever? How can this be addressed?

Biologists want the structure of the protein, not just the positions
of some atoms in some crystal. They want to know where atom N-delta in
the sidechain of residue LEU42 in chain A of the protein is.

 > > Right, that's where H5MD beats the PDB format (and also mmCIF) - but
 > > the PDB doesn't care about that. They have no time-dependent data.
 > 
 > All the PDB-abusers around would care, though.

Right. H5MD is of interest to PDB-abusers, but not so much to the PDB.
Except perhaps to get rid of the abusers ;-)

Konrad.
-- 
---------------------------------------------------------------------
Konrad Hinsen
Centre de Biophysique Moléculaire, CNRS Orléans
Synchrotron Soleil - Division Expériences
Saint Aubin - BP 48
91192 Gif sur Yvette Cedex, France
Tel. +33-1 69 35 97 15
E-Mail: research AT khinsen DOT fastmail DOT net
http://dirac.cnrs-orleans.fr/~hinsen/
ORCID: http://orcid.org/0000-0003-0330-9428
Twitter: @khinsen
---------------------------------------------------------------------