Forums

I should point out that there already exists a huge library of
additional functionality for manipulating SBML models particularly with
regard to model analysis in SBW. We are currently testing for example a
new algorithm for computing conservation laws and model reduction which
avoids the use to gaussian reduction which may be unstable for large
systems.

Herbert Sauro

-----Original Message-----
From: Karthik Raman [mailto:karthik@rishi.serc.iisc.ernet.in]
Sent: Wednesday, July 06, 2005 7:07 PM
To: SBML Discussion List
Subject: Re: [sbml-discuss] What do SBML libraries need to focus on
next?

Hi Nick

I think Ben does have some of these in mind. In the Hackathon at Tokyo,
I had indeed discussed with him about people writing some add-on codes
to libSBML, such as extracting a stoichometric matrix, that would be of
great use (there's an example working code for this on the
libsbml-discuss web interface).

It would be great if this and other such routine model tasks would be
integrated into libSBML.

Regards

Karthik

********************************************************

KARTHIK RAMAN
Graduate Research Student
Supercomputer Education and Research Centre
Indian Institute of Science
Bangalore - 560 012

Phone: +91-80-22932469 (Ext 24)
+91-80-23601409 (Ext 24)
+91-98862-22404 (Mobile)

E-mail : karthik AT rishi.serc.iisc.ernet.in
Webpage : http://rishi.serc.iisc.ernet.in/~karthik/
Blog : karthikraman.blogspot.com

*******************************************************


On Wed, 6 Jul 2005, Nicholas Allen wrote:

>
> It's been roughly 3 years since the idea of SBML L2 was proposed.
Since that time, I think that the underlying SBML libraries have gotten
quite good at the mechanics of reading and writing SBML. There's still
a lot of work left to be done to actually understand and use SBML, but
that task is shared with SBML applications. So, I thought I'd ask
everyone for their opinions on what SBML libraries should do to help
make that happen, and what support is most important. I've come up with
a few ways that SBML libraries could support applications better and I'd
like to know about other ideas people have.
>
> - Validating models for semantic correctness. libSBML already can do
a number of consistency checks on a model mentioned in the L2
specification, but there are some specification requirements left
unchecked and it can be difficult to get feedback about what needs to be
changed to make a model valid. JigCell has a smaller number of
consistency checks, and I'd like to improve on this.
>
> - Translating id references in the model to the actual definition that
is being referenced. References in an SBML model give an identifier,
and it's up to you to figure out what that identifier actually means.
It can be tricky to get this correct because of the way default units
work and namespace clashes. I've been adding a number of methods to
JigCell to resolve identifiers and I know Ben has mentioned doing this
for libSBML as well.
>
> - Support for unit consistency checking and conversion. I know this
is a big one as many SBML applications do a pretty poor job handling
units at the current time. JigCell does the bare minimum to handle
units and leaves it up to applications deal with the problem currently.
>
> - Tools for working with MathML. I think that libSBML has facilities
for performing infix<->MathML conversions and JigCell does too, but it's
still quite hard to interpret and manipulate MathML expressions after
the SBML library has gone over the model.
>
> - Computing higher-level information about a model. There's a lot of
duplicated code for computing stoichiometry matrices, differential
equations, moiety conservation relations, and the like from a model.
SBML libraries could provide these services so that people don't have to
write their own. I'm thinking about moving a number of these
computations from JigCell applications to the SBML library so that
they're available to users.
>
> -- Nick Allen
> nallen@vt.edu
>
>