[Molecularmechanics] first steps

[Konrad Hinsen]

On Tuesday 21 October 2003 10:00, Peter Murray-Rust wrote:

>   * There is a lot of overlap with other computational molecular areas,
> especially crystal and material calculations. I believe that the
> abstractions of force field, trajectories, etc. are common and we may have
> useful interaction with non-bio people

Definitely.

>   * There will be a need for generic libraries (esp i/o for file formats,
> XML and HCF/NETCDF) and again this overlaps well with these other
> communities

Right.

> I may therefore point some of my collaborators in these fields to your
> pages.

That sounds very useful.  The non-bio people are not so much organized around 
big codes (it is more usual to work with one's own code), so they are not so 
easy to identify.

> libraries. Are we agreed that F90 is taken as a universally available
> technology?

Partially. I think it is safe to assume that anyone doing Fortran development 
today has an F90 compiler. On the other hand, there are people who move away 
from Fortran just because the compilers are not universally (read: freely) 
available.

> What is the position of AMBER and CHARMM as they are not open source? My
> current approach would required wrapping them with input adapters (probably

We have Michael Crowley on board who works on both. In general, I think we 
should leave implementation as much as possible to the developers of the 
respective codes. I think we have enough critical mass with the developers 
who participate that we do not have to adapt "foreign" codes to whatever 
format we agree on.

> CMLComp (CML for computational chemistry) has addressed most of these but

Is that the same as CCML?

> crystal mechanics program GULP). I am keen to explore the distribution of
> machine-understandable (parsable) functional forms so that these do not
> necessarily have to be hardcoded. MathML may be useful for this.

That sounds like an interesting proposition, but I think the problem of force 
field representation, at least for biomolecules is elsewhere. The functional 
forms of the individual terms are sufficiently simple, but the rules for 
determining which terms are required to describe a particular molecule are 
quite complex and very different from one forcefield to another. They involve 
both algorithmic rules and explicit lists for each specific amino acid 
residue.

> I think it's important to welcome developers from non-Open environments.
> For example someone working on a commercial code might have important input
> to an F90 library.

Definitely. The only condition should be that the resulting standards are free 
and freely useable, which would exclude for example the use of patented 
technology.

Konrad.
-- 
-------------------------------------------------------------------------------
Konrad Hinsen                            | E-Mail: hinsen@cnrs-orleans.fr
Centre de Biophysique Moleculaire (CNRS) | Tel.: +33-2.38.25.56.24
Rue Charles Sadron                       | Fax:  +33-2.38.63.15.17
45071 Orleans Cedex 2                    | Deutsch/Esperanto/English/
France                                   | Nederlands/Francais
-------------------------------------------------------------------------------