Another reason for inactivity is perhaps that at various times we are too
busy to post. At the moment the systems biology team at caltech (of which I
am a member) are busy developing a release for the systems biology workbench
(SBW) for November the 1st 2001 in time for the system biology conference
this coming November, I hope many readers of this group will try to attend.
Details of the SBW can be found at www.cds.caltech.edu/erato The site is
about to be revamped, most likely this coming week with new documentation
and a new look. For those who are unaware of the SBW here is a short
description.
The project has gone through a number of phases but is now settled on the
following goal.
As many of you know, there are numerous simulators and other computational
resourses 'out there' which are useful to readers of this group. However one
of the problems is that simulations or models
built in one package cannot be transfered to another package. So if you want
to use some feature which is not supported in the package you're currently
using, it is not possible to move your model easily to an alternative
package which has the required capability.
To get round this problem we have two approaches:
1) Develop a community based exchange format which will allow users to swap
models from one package to another. The first level of this was released on
March 1, called SBML Level 1, that is systems biololgy markup language,
level 1.
2) The second and more ambitious task we are attempting is to develop the
so-called systems biology workbench. This is not a simulator or
computational resourse in itself but a means for allowing one resource to
talk directly to another resource. One of the ideas for SBW is that it is
distributable, that is you will be able to use resources across the
internet.
There are of course existing technologies for sharing resouces, eg CORBA,
RPC etc. However we felt that these tended to be over engineered and far too
complex for the general biological computational community to work with
unless one had significant grant support to pay for expensive programmers.
With inspiriation from the new peer to peer technologies that now abound (eg
Naspter, Gnuella), we have devised what I believe to be a simple,
lightweight but very effective way of sharing resources. From playing around
with a prototype I found it remarkably easy to make an application (of
course you must have access to the sources) SBW aware, for example we took a
Gillespie stochastic simulator from a colleague at the University and made
it SBW ready in litteraly only half to one hour. Once the Gillespie apps
made SBW aware, everything that was connected to SBW could make use of it.
The critical feature of SBW is that it is very straight forward to turn a
non-SBW complient application into one which is. Initial tests indicate that
our design has good data transfer performance characteristics so that it is
suitable for computational type work. It is also language neutral so that
users can develop SBW complient applications using their faviourite language
(Slated for November, C, C++, Java, Python). In the intial release SBW will
run on Windows and Linux platforms, a Mac version depends on the new Mac OS
just released but it's a possiblity we will look into at a later date.
Having seen a prototype SBW in action and I am quite excited by its
potential. If the two programmers, Andrew Finney and Mike Huckacan deliver
what is promised then I will definitely make Jarnac, JDesginer and anything
else I write SBW compatible. If you can get travel support I would
definitely make an effort to come to the systems biology confernece this
coming November where SBW will be released and demonstrated.
Herbert Sauro
Caltech
