| Author | Topic |
|---|
Posts: 34
Registered:
May 2004
|
|
Little b
|
10 Jul '05 14:14
|
 |
|
This is from John Russell's column in Bio-IT World last week.
Two questions: Jeremy, do these high profile systems biologists know about
SBML and the array of solvers that support it? Everyone else, why LISP?
Harvard Tackles Systems Biology
Towards the end of this summer, Jeremy Gunawardena and Aneil Mallavarapu of
the Virtual Cell Program at Harvard Medical School's Department of Systems
Biology expect to unleash b -- pronounced "little b" -- a new open-source
computer language they hope will energize the biological modeling community
as nothing has before.
Today, a handful of companies and academic groups are building computational
models of disease and biological pathways. To a considerable extent, these
scattered efforts remain in their infancy (although modeling has recently
gained traction inside a few biopharmas). Moreover, the companies charge for
their services and tools and have a vested interest in controlling their
proprietary technology. On balance, there has been little synergy between
these isolated pockets of progress.
Imagine instead a biologist-friendly language designed specifically to
encapsulate biological knowledge and constantly pounded on and improved upon
by an open-source community. Add to this vision a growing library of models
(and model fragments) written in b and contributed by researchers worldwide
for free use by other researchers. Based on list processing (LISP), b is
designed to be modular, emphasizing the reusability of modules and a
Lego-like approach to model building.
Currently, b focuses on biochemical modeling using ordinary differential
equations. However, Mallavarapu says, "We want to expand to describe other
types of models like partial differential equations, and we're working on
what it will take to write stochastic models."
It's not a great leap to conceive of a GenBank for computational models that
researchers could contribute to and draw from. Imagine the interesting and
productive in silico research that an army of modelers might undertake.
This is a grand vision for a language with such a diminutive name, and it is
one of the first concrete projects emerging from Harvard's ambitious foray
into systems biology. The department is a little more than a year old, has
enrolled nine students in its Ph.D. program, and has assembled an impressive
faculty led by chair Marc Kirschner, deputy chair Timothy Mitchison, and
Lewis Cantley.
They were "the gang of three who believed sufficiently that something new
was happening to come together and make this [department] happen," says
Gunawardena, who is the director of the Virtual Cell Program.
Language Barriers
A self-professed "mathematician who fell from grace," Gunawardena had stints
in academia and industry, including a long stay at Hewlett-Packard doing
industrial research. While at HP, he caught the biology bug and moved to
Harvard's Bauer Center for Genomic Research. That was his first opportunity,
he says, "to actually live among biologists and understand what was going
on. Pretty much everything I thought before I came [to Bauer] turned out to
be...modified substantially." When ideas for the new systems biology
department bubbled up, he was invited to join.
Mallavarapu is a cell biologist and biochemist by training. He took his
Ph.D. with Mitchison at the University of California, San Francisco, working
on photomarking technologies to visualize cytoskeletal dynamics. During the
past few years, Mallavarapu worked at Millennium Pharmaceuticals developing
technology, writing software, and thinking about what is now b. He credits
conversations with Gunawardena for stoking his desire to make the language a
reality, and he recently joined the Virtual Cell Program as a research
scientist. These guys have great jobs!
Certainly, tricky issues remain. Defining b so that it can be easily
compiled on various LISP platforms isn't trivial. Attracting an early
adopter community will be important. Writing an easy-to-use graphical user
interface is another priority, as is getting models and modeling techniques
published in peer-reviewed literature.
"I think [it's] the classical catch-22 problem: Until there's a community of
people speaking [a language], why would you want to learn it?" Gunawardena
says. "I'm very keen on little b in the context of some courses that we'll
be teaching [and] think that's going to be a very influential early adopter
community. I think it's likely that some of our colleagues at the medical
school and also at MIT are very interested."
It will be interesting to watch how big little b becomes.
|
|
|
Posts: 46
Registered:
September 2003
|
|
Re: Little b
|
11 Jul '05 01:02
|
|
|
Hi Robert,
Funny you should ask. I met with Aneil Mallavarapu last week to
discuss writing the SBML module for little b. We have the choice of
Martin Ginkel's libsbml lisp bindings and Marco Antoniotti's pure lisp
implementation. The export utility will be easy. What will be harder
is inferring what kinetic rate law assumptions were made when importing
SBML. We have some ideas on how to do this, and we are paying close
attention to the recent discussions on controlled vocabularies for
kinetic rate laws.
As for why lisp is useful for writing biological programming languages,
I think Jeff Shrager said it pretty well:
http://nostoc.stanford.edu/Docs/whylisp.html
Aneil is a pragmatist. He started out trying to implement little b in
conventional programming languages but kept running into limitations
that eventually drove him to lisp.
But I'll let him speak for himself.
Sincerely,
Jeremy
On Sun, 2005-07-10 at 17:14 -0400, rphair@integrativebioinformatics.com
wrote:
> This is from John Russell’s column in Bio-IT World last week.
>
>
>
> Two questions: Jeremy, do these high profile systems biologists know
> about SBML and the array of solvers that support it? Everyone else,
> why LISP?
>
>
>
> Harvard Tackles Systems Biology
>
> Towards the end of this summer, Jeremy Gunawardena and Aneil
> Mallavarapu of the Virtual Cell Program at Harvard Medical School's
> Department of Systems Biology expect to unleash b -- pronounced
> "little b" -- a new open-source computer language they hope will
> energize the biological modeling community as nothing has before.
>
>
>
> Today, a handful of companies and academic groups are building
> computational models of disease and biological pathways. To a
> considerable extent, these scattered efforts remain in their infancy
> (although modeling has recently gained traction inside a few
> biopharmas). Moreover, the companies charge for their services and
> tools and have a vested interest in controlling their proprietary
> technology. On balance, there has been little synergy between these
> isolated pockets of progress.
>
>
>
> Imagine instead a biologist-friendly language designed specifically to
> encapsulate biological knowledge and constantly pounded on and
> improved upon by an open-source community. Add to this vision a
> growing library of models (and model fragments) written in b and
> contributed by researchers worldwide for free use by other
> researchers. Based on list processing (LISP), b is designed to be
> modular, emphasizing the reusability of modules and a Lego-like
> approach to model building.
>
>
>
> Currently, b focuses on biochemical modeling using ordinary
> differential equations. However, Mallavarapu says, "We want to expand
> to describe other types of models like partial differential equations,
> and we're working on what it will take to write stochastic models."
>
> It's not a great leap to conceive of a GenBank for computational
> models that researchers could contribute to and draw from. Imagine the
> interesting and productive in silico research that an army of modelers
> might undertake.
>
>
>
> This is a grand vision for a language with such a diminutive name, and
> it is one of the first concrete projects emerging from Harvard's
> ambitious foray into systems biology. The department is a little more
> than a year old, has enrolled nine students in its Ph.D. program, and
> has assembled an impressive faculty led by chair Marc Kirschner,
> deputy chair Timothy Mitchison, and Lewis Cantley.
>
>
>
> They were "the gang of three who believed sufficiently that something
> new was happening to come together and make this [department] happen,"
> says Gunawardena, who is the director of the Virtual Cell Program.
>
>
>
> Language Barriers
>
> A self-professed "mathematician who fell from grace," Gunawardena had
> stints in academia and industry, including a long stay at Hewlett-
> Packard doing industrial research. While at HP, he caught the biology
> bug and moved to Harvard's Bauer Center for Genomic Research. That was
> his first opportunity, he says, "to actually live among biologists and
> understand what was going on. Pretty much everything I thought before
> I came [to Bauer] turned out to be...modified substantially." When
> ideas for the new systems biology department bubbled up, he was
> invited to join.
>
>
>
> Mallavarapu is a cell biologist and biochemist by training. He took
> his Ph.D. with Mitchison at the University of California, San
> Francisco, working on photomarking technologies to visualize
> cytoskeletal dynamics. During the past few years, Mallavarapu worked
> at Millennium Pharmaceuticals developing technology, writing software,
> and thinking about what is now b. He credits conversations with
> Gunawardena for stoking his desire to make the language a reality, and
> he recently joined the Virtual Cell Program as a research scientist.
> These guys have great jobs!
>
>
>
> Certainly, tricky issues remain. Defining b so that it can be easily
> compiled on various LISP platforms isn't trivial. Attracting an early
> adopter community will be important. Writing an easy-to-use graphical
> user interface is another priority, as is getting models and modeling
> techniques published in peer-reviewed literature.
>
>
>
> "I think [it's] the classical catch-22 problem: Until there's a
> community of people speaking [a language], why would you want to learn
> it?" Gunawardena says. "I'm very keen on little b in the context of
> some courses that we'll be teaching [and] think that's going to be a
> very influential early adopter community. I think it's likely that
> some of our colleagues at the medical school and also at MIT are very
> interested."
>
>
>
> It will be interesting to watch how big little b becomes.
>
>
>
>
>
>
|
|
|
Posts: 349
Registered:
September 2003
|
|
RE: Little b
|
11 Jul '05 12:53
|
|
|
I saw this article on friday and I thought it was very biased. For
example the statement, "It's not a great leap to conceive of a GenBank
for computational models that researchers could contribute to and draw
from." reads as if no such thing currently exists, have they not heard
of SigPath or biomodels.net (which was mentioned in a recent nature
editorial). I know the people in this group and they know full well what
is going on in the community, I think it was just a cheap publicity
stunt to get 'b' marketed. In any case 'b' is not the first of it kind,
there are a number of existing projects which have a similar objective.
Herbert Sauro
_____
From: rphair@integrativebioinformatics.com
[mailto:rphair@integrativebioinformatics.com]
Sent: Sunday, July 10, 2005 1:14 PM
To: 'SBML Discussion List'
Subject: [sbml-discuss] Little b
This is from John Russell's column in Bio-IT World last week.
Two questions: Jeremy, do these high profile systems biologists know
about SBML and the array of solvers that support it? Everyone else, why
LISP?
Harvard Tackles Systems Biology
Towards the end of this summer, Jeremy Gunawardena and Aneil Mallavarapu
of the Virtual Cell Program at Harvard Medical School's Department of
Systems Biology expect to unleash b -- pronounced "little b" -- a new
open-source computer language they hope will energize the biological
modeling community as nothing has before.
Today, a handful of companies and academic groups are building
computational models of disease and biological pathways. To a
considerable extent, these scattered efforts remain in their infancy
(although modeling has recently gained traction inside a few
biopharmas). Moreover, the companies charge for their services and tools
and have a vested interest in controlling their proprietary technology.
On balance, there has been little synergy between these isolated pockets
of progress.
Imagine instead a biologist-friendly language designed specifically to
encapsulate biological knowledge and constantly pounded on and improved
upon by an open-source community. Add to this vision a growing library
of models (and model fragments) written in b and contributed by
researchers worldwide for free use by other researchers. Based on list
processing (LISP), b is designed to be modular, emphasizing the
reusability of modules and a Lego-like approach to model building.
Currently, b focuses on biochemical modeling using ordinary differential
equations. However, Mallavarapu says, "We want to expand to describe
other types of models like partial differential equations, and we're
working on what it will take to write stochastic models."
It's not a great leap to conceive of a GenBank for computational models
that researchers could contribute to and draw from. Imagine the
interesting and productive in silico research that an army of modelers
might undertake.
This is a grand vision for a language with such a diminutive name, and
it is one of the first concrete projects emerging from Harvard's
ambitious foray into systems biology. The department is a little more
than a year old, has enrolled nine students in its Ph.D. program, and
has assembled an impressive faculty led by chair Marc Kirschner, deputy
chair Timothy Mitchison, and Lewis Cantley.
They were "the gang of three who believed sufficiently that something
new was happening to come together and make this [department] happen,"
says Gunawardena, who is the director of the Virtual Cell Program.
Language Barriers
A self-professed "mathematician who fell from grace," Gunawardena had
stints in academia and industry, including a long stay at
Hewlett-Packard doing industrial research. While at HP, he caught the
biology bug and moved to Harvard's Bauer Center for Genomic Research.
That was his first opportunity, he says, "to actually live among
biologists and understand what was going on. Pretty much everything I
thought before I came [to Bauer] turned out to be...modified
substantially." When ideas for the new systems biology department
bubbled up, he was invited to join.
Mallavarapu is a cell biologist and biochemist by training. He took his
Ph.D. with Mitchison at the University of California, San Francisco,
working on photomarking technologies to visualize cytoskeletal dynamics.
During the past few years, Mallavarapu worked at Millennium
Pharmaceuticals developing technology, writing software, and thinking
about what is now b. He credits conversations with Gunawardena for
stoking his desire to make the language a reality, and he recently
joined the Virtual Cell Program as a research scientist. These guys have
great jobs!
Certainly, tricky issues remain. Defining b so that it can be easily
compiled on various LISP platforms isn't trivial. Attracting an early
adopter community will be important. Writing an easy-to-use graphical
user interface is another priority, as is getting models and modeling
techniques published in peer-reviewed literature.
"I think [it's] the classical catch-22 problem: Until there's a
community of people speaking [a language], why would you want to learn
it?" Gunawardena says. "I'm very keen on little b in the context of some
courses that we'll be teaching [and] think that's going to be a very
influential early adopter community. I think it's likely that some of
our colleagues at the medical school and also at MIT are very
interested."
It will be interesting to watch how big little b becomes.
|
|
|
Posts: 349
Registered:
September 2003
|
|
RE: Little b
|
11 Jul '05 12:59
|
|
|
Jermey, I looked at the Lisp article you cited, I have to say I don't
agree. You yourself are quite at home with computing as are many here on
this group and naturally a Lisp approach is very attractive to computer
folk, however in the biology community it is a different matter. Most
biologists are not computer savy enough to be able to takle Lisp or most
computer languages for that matter, and frankly I don't think they
should have to either. They have enough on their plate as it is without
making them jump through hoops to get simple jobs done. Matlab is about
the limit and even then it is a small struggle to teach people how to
use it.
Herbert Sauro
-----Original Message-----
From: Jeremy Zucker [mailto:zucker@research.dfci.harvard.edu]
Sent: Monday, July 11, 2005 12:03 AM
To: SBML Discussion List; mallavar@hms.harvard.edu
Subject: Re: [sbml-discuss] Little b
Hi Robert,
Funny you should ask. I met with Aneil Mallavarapu last week to
discuss writing the SBML module for little b. We have the choice of
Martin Ginkel's libsbml lisp bindings and Marco Antoniotti's pure lisp
implementation. The export utility will be easy. What will be harder
is inferring what kinetic rate law assumptions were made when importing
SBML. We have some ideas on how to do this, and we are paying close
attention to the recent discussions on controlled vocabularies for
kinetic rate laws.
As for why lisp is useful for writing biological programming languages,
I think Jeff Shrager said it pretty well:
http://nostoc.stanford.edu/Docs/whylisp.html
Aneil is a pragmatist. He started out trying to implement little b in
conventional programming languages but kept running into limitations
that eventually drove him to lisp.
But I'll let him speak for himself.
Sincerely,
Jeremy
On Sun, 2005-07-10 at 17:14 -0400, rphair@integrativebioinformatics.com
wrote:
> This is from John Russell's column in Bio-IT World last week.
>
>
>
> Two questions: Jeremy, do these high profile systems biologists know
> about SBML and the array of solvers that support it? Everyone else,
> why LISP?
>
>
>
> Harvard Tackles Systems Biology
>
> Towards the end of this summer, Jeremy Gunawardena and Aneil
> Mallavarapu of the Virtual Cell Program at Harvard Medical School's
> Department of Systems Biology expect to unleash b -- pronounced
> "little b" -- a new open-source computer language they hope will
> energize the biological modeling community as nothing has before.
>
>
>
> Today, a handful of companies and academic groups are building
> computational models of disease and biological pathways. To a
> considerable extent, these scattered efforts remain in their infancy
> (although modeling has recently gained traction inside a few
> biopharmas). Moreover, the companies charge for their services and
> tools and have a vested interest in controlling their proprietary
> technology. On balance, there has been little synergy between these
> isolated pockets of progress.
>
>
>
> Imagine instead a biologist-friendly language designed specifically to
> encapsulate biological knowledge and constantly pounded on and
> improved upon by an open-source community. Add to this vision a
> growing library of models (and model fragments) written in b and
> contributed by researchers worldwide for free use by other
> researchers. Based on list processing (LISP), b is designed to be
> modular, emphasizing the reusability of modules and a Lego-like
> approach to model building.
>
>
>
> Currently, b focuses on biochemical modeling using ordinary
> differential equations. However, Mallavarapu says, "We want to expand
> to describe other types of models like partial differential equations,
> and we're working on what it will take to write stochastic models."
>
> It's not a great leap to conceive of a GenBank for computational
> models that researchers could contribute to and draw from. Imagine the
> interesting and productive in silico research that an army of modelers
> might undertake.
>
>
>
> This is a grand vision for a language with such a diminutive name, and
> it is one of the first concrete projects emerging from Harvard's
> ambitious foray into systems biology. The department is a little more
> than a year old, has enrolled nine students in its Ph.D. program, and
> has assembled an impressive faculty led by chair Marc Kirschner,
> deputy chair Timothy Mitchison, and Lewis Cantley.
>
>
>
> They were "the gang of three who believed sufficiently that something
> new was happening to come together and make this [department] happen,"
> says Gunawardena, who is the director of the Virtual Cell Program.
>
>
>
> Language Barriers
>
> A self-professed "mathematician who fell from grace," Gunawardena had
> stints in academia and industry, including a long stay at Hewlett-
> Packard doing industrial research. While at HP, he caught the biology
> bug and moved to Harvard's Bauer Center for Genomic Research. That was
> his first opportunity, he says, "to actually live among biologists and
> understand what was going on. Pretty much everything I thought before
> I came [to Bauer] turned out to be...modified substantially." When
> ideas for the new systems biology department bubbled up, he was
> invited to join.
>
>
>
> Mallavarapu is a cell biologist and biochemist by training. He took
> his Ph.D. with Mitchison at the University of California, San
> Francisco, working on photomarking technologies to visualize
> cytoskeletal dynamics. During the past few years, Mallavarapu worked
> at Millennium Pharmaceuticals developing technology, writing software,
> and thinking about what is now b. He credits conversations with
> Gunawardena for stoking his desire to make the language a reality, and
> he recently joined the Virtual Cell Program as a research scientist.
> These guys have great jobs!
>
>
>
> Certainly, tricky issues remain. Defining b so that it can be easily
> compiled on various LISP platforms isn't trivial. Attracting an early
> adopter community will be important. Writing an easy-to-use graphical
> user interface is another priority, as is getting models and modeling
> techniques published in peer-reviewed literature.
>
>
>
> "I think [it's] the classical catch-22 problem: Until there's a
> community of people speaking [a language], why would you want to learn
> it?" Gunawardena says. "I'm very keen on little b in the context of
> some courses that we'll be teaching [and] think that's going to be a
> very influential early adopter community. I think it's likely that
> some of our colleagues at the medical school and also at MIT are very
> interested."
>
>
>
> It will be interesting to watch how big little b becomes.
>
>
>
>
>
>
|
|
|
Posts: 961
Registered:
October 2003
|
|
|
Posts: 16
Registered:
April 2004
|
|
Re: Little b
|
13 Jul '05 07:35
|
|
|
Hi
Yep. I agree. I always discount announcements made in the form of
"press releases". The job of a p.r. person (or of a person who
temporarily puts on such a hat) is not that of presenting "evidence",
but to "sell" a product.
I would just check the system and see what it can really do.
Cheers
--
Marco
On Jul 12, 2005, at 10:34 PM, Michael Hucka wrote:
> I was also somewhat surprised at the content of that
> announcement (I, too, know this group), but I'm chalking up
> the mild overstatements to a "press release effect" :-). We
> learned this hard lesson with our BioModels Database release
> announcement earlier this year. As people saw, we managed
> to tick off collaborators unexpectedly.
>
> There's something about press releases....
>
> MH
>
>
--
Marco Antoniotti http://bioinformatics.nyu.edu
NYU Courant Bioinformatics Group tel. +1 - 212 - 998 3488
715 Broadway 10th FL fax. +1 - 212 - 998 3484
New York, NY, 10003, U.S.A.
|
|
|
Posts: 1
Registered:
July 2005
|
|
RE: Little b
|
14 Jul '05 06:29
|
|
|
Hi Jeremy (and Robert)-
Thanks for the forward. We spoke at some length during the interview about
SBML. It's strange that the proprietary tools got a mention but SBML didn't.
I think the bit about "energizing the biological modeling community like
nothing ever before" should be regarded as a bit of journalistic art.
Anyway, I've sketched out some thoughts about the relationship between b and
SBML on the website (http://www.littleb.org/faq.html#compare-with-xml). Let
me know what you think.
Why LISP? LISP is a wonderfully flexible language designed for symbolic
reasoning. And model building is largely about symbolic reasoning. There are
other candidates - ML, Prolog, Haskell - but I think LISP wins the day in
terms of its flexibility, particularly the ability to extend the language
with macros. In essence, macros allow you to define sublanguages within
LISP. I'm not entirely certain what is the exactly right combination of
features in an "ideal" modeling language - Prolog-style inference,
first-order functions, object-oriented programming, subgraph isomorphism
detection - so the fact that LISP allows you to build these features into it
with macros seems to be a big plus. This is what drove me to it, in any
case. There's more to be said on this topic, but I'm currently on vacation,
so I'll have to leave this for another time.
Cheers,
Aneil
PS: Sorry for the late response - there were some email snafus along the
way & I'm only intermittently connected to the net.
|
|
|
Posts: 469
Registered:
October 2003
|
|
RE: Little b
|
14 Jul '05 11:46
|
|
|
"In little b, by contrast [to XML], the "specifications" are written
directly in b, rather than English"
But this is exactly what XML schema is. XML schema is an XML language
specifying other XML languages. And although the SBML schema stayed simple
so far, we could transfer quite a lot of the spec directly in the schema,
for instance using keys.
(I don't want to start a flame. I just think XML possibilities are often
underated, particularly in SBML community.)
--
Nicolas LE NOVČRE, Computational Neurobiology,
EMBL-EBI, Wellcome-Trust Genome Campus, Hinxton, Cambridge, CB10 1SD, UK
Tel: +44(0)1223 494 521, Fax: +44(0)1223 494 468, Mob: +33(0)689218676
http://www.ebi.ac.uk/~lenov
|
|
|
Posts: 349
Registered:
September 2003
|
|
RE: Little b
|
14 Jul '05 14:06
|
|
|
The trouble with many languages is that most are written by computer guys for computer guys, one could argue even further that XML is written by computers for computers :) (any non-computer specialist who has done significant SBML Level 2 editing will know what I mean). SBML is great for model interchange, no question about it, for model representation for humans it is a different matter. We of course use jarnac which is currently adequate but it isn't at all very good at composition. Other people like a visual representation, hence we wrote jdesigner, others like a more logical represention such as Simpathica, each has it's strengths and weaknesses. At the moment we know very little about b other than it's in Lisp, I sure it will be great, but it is one of many representations that currently exist.
Herbert
-----Original Message-----
From: Nicolas Le Novčre [mailto:lenov@ebi.ac.uk]
Sent: Thursday, July 14, 2005 10:46 AM
To: SBML Discussion List
Subject: RE: [sbml-discuss] Little b
"In little b, by contrast [to XML], the "specifications" are written directly in b, rather than English"
But this is exactly what XML schema is. XML schema is an XML language specifying other XML languages. And although the SBML schema stayed simple so far, we could transfer quite a lot of the spec directly in the schema, for instance using keys.
(I don't want to start a flame. I just think XML possibilities are often underated, particularly in SBML community.)
--
Nicolas LE NOVČRE, Computational Neurobiology, EMBL-EBI, Wellcome-Trust Genome Campus, Hinxton, Cambridge, CB10 1SD, UK
Tel: +44(0)1223 494 521, Fax: +44(0)1223 494 468, Mob: +33(0)689218676 http://www.ebi.ac.uk/~lenov
|
|
|
Posts: 34
Registered:
May 2004
|
|
RE: Little b
|
15 Jul '05 11:15
|
|
|
Aneil-
Thanks for responding so completely to my query about Lisp. I've imagined
you were not planning to ask biologists to learn Common Lisp or even Scheme
but rather that you chose Lisp because it makes some difficult programming
chores easy to code or makes some impossible things feasible. The links to
ViaWeb/Yahoo Stores were inspiring , but I wonder what functionality you are
thinking you will be able to offer in Little b that you simply could not
code in C++ or Java. Are you thinking of macros that speak the language of
MCA or flux balance analysis?
Also, are you planning a native Little b file format that other tools could
read/write, or are you planning to uses SBML with annotations?
Regards,
Robert D Phair, PhD
Chief Scientific Officer
Integrative BioInformatics Inc
Rockville, MD 20854 USA
301.315.8114
_____
From: sbml-discuss-bounces@caltech.edu
[mailto:sbml-discuss-bounces@caltech.edu] On Behalf Of Aneil Mallavarapu
Sent: Thursday, July 14, 2005 9:30 AM
To: Jeremy Zucker; sbml-discuss@caltech.edu
Cc: Jeremy Gunawardena
Subject: RE: [sbml-discuss] Little b
Hi Jeremy (and Robert)-
Thanks for the forward. We spoke at some length during the interview about
SBML. It's strange that the proprietary tools got a mention but SBML
didn't. I think the bit about "energizing the biological modeling community
like nothing ever before" should be regarded as a bit of journalistic art.
Anyway, I've sketched out some thoughts about the relationship between b and
SBML on the website ( http://www.littleb.org/faq.html#compare-with-xml).
Let me know what you think.
Why LISP? LISP is a wonderfully flexible language designed for symbolic
reasoning. And model building is largely about symbolic reasoning. There
are other candidates - ML, Prolog, Haskell - but I think LISP wins the day
in terms of its flexibility, particularly the ability to extend the language
with macros. In essence, macros allow you to define sublanguages within
LISP. I'm not entirely certain what is the exactly right combination of
features in an "ideal" modeling language - Prolog-style inference,
first-order functions, object-oriented programming, subgraph isomorphism
detection - so the fact that LISP allows you to build these features into it
with macros seems to be a big plus. This is what drove me to it, in any
case. There's more to be said on this topic, but I'm currently on vacation,
so I'll have to leave this for another time.
Cheers,
Aneil
PS: Sorry for the late response - there were some email snafus along the way
& I'm only intermittently connected to the net.
|
|
|
Posts: 4
Location:
Harvard University
Registered:
July 2005
|
|
|
Posts: 4
Location:
Harvard University
Registered:
July 2005
|
|
|
Posts: 4
Location:
Harvard University
Registered:
July 2005
|
|
Re: RE: Little b
|
26 Jul '05 09:58
|
|
|
Hi Nicholas -
In little b, what we're trying to do is specify how a particular description at one "level" (e.g., biology) should be translated into another level (e.g., mathematics). In general, we'll need to define functions and logical rules which accomplish these transformations automatically. The idea is to do automated model generation, not just validation. We'd also like to be able to swap the rules so that a particular description of biology can be transformed to one or more mathematical representations. I haven't really kept up with the evolution of XML schema the past few years, but my sense is that they're not really up to this kind of complex task.
Nicolas Le Novčre wrote:
"In little b, by contrast [to XML], the "specifications" are written
directly in b, rather than English"
But this is exactly what XML schema is. XML schema is an XML language specifying other XML languages. And although the SBML schema stayed simple so far, we could transfer quite a lot of the spec directly in the schema, for instance using keys.
|
|
|
Posts: 4
Location:
Harvard University
Registered:
July 2005
|
|
Re: RE: Little b
|
26 Jul '05 10:57
|
|
|
>I wonder what functionality you are
>thinking you will be able to offer in Little b that you simply
>could not code in C++ or Java. Are you thinking of macros that
>speak the language of MCA or flux balance analysis?
C++ and Java are turing-complete, and just as expressive as LISP from that mathematical perspective, but so is Assembly language. What LISP enabled me to do was define a terse syntax for describing objects, logical rules and symbolic mathematical expressions. LISP enabled me to do this in a way that was readable, writable. Without the syntactic & semantic capabilities, I wouldn't have been able to manage the complexity of the project.
To get a sense of what's going on, take a look at a recent presentation (http://www.littleb.org/presentation.ppt). It's hard to imagine retrofitting C++ or Java with logic-programming, symbolic math and new syntax to support these without rewriting the compilers. In LISP, this achieved by extending the language with macros (a faq on the little b site discusses macros briefly).
(Depending on the requirements, macros for FBA & MCA might be useful, though I imagine ordinary functions would fit the bill. Macros should be used sparingly.)
>Also, are you planning a native Little b file format that other
>tools could read/write, or are you planning to uses SBML with
>annotations?
On the output side, we're planning to write SBML with annotations. To get things up and running, we've written a quick-and-dirty MatLab converter.
For input (into b), other tools could write little b code directly - but XML might be a better format for interchange between programs: SBML with annotations (or extended tags) might fit the bill.
Aneil
|
|
|
Posts: 961
Registered:
October 2003
|
|
|
Posts: 349
Registered:
September 2003
|
|
|
Posts: 15
Registered:
December 2004
|
|
|
Posts: 16
Registered:
April 2004
|
|
Re: Re: RE: Little b
|
29 Jul '05 09:27
|
|
|
Well, given that they learn to program in R (and Perl and Python and
Ruby and Tcl and SLDJ), they can also learn Lisp.
Cheers
--
Marco
On Jul 29, 2005, at 11:33 AM, Tomas Radivoyevitch wrote:
> Backing up what Herbert just said, biologists and MDs who took my
> stats class did indeed learn how to program in R.
>
>
> ----- Original Message ----- From: "Herbert Sauro"
> <Herbert_Sauro@kgi.edu>
> To: "'SBML Discussion List'" <sbml-discuss@caltech.edu>
> Sent: Wednesday, July 27, 2005 4:23 PM
> Subject: RE: [sbml-discuss] Re: RE: Little b
>
>
>>
>>> The "unnatural" structure of computer languages comes from the need
>>> for
>> precise
>>> unambiguous expression. If models are to become part of mainstream
>>> of
>> biology,
>>> perhaps it will be important for biologists to express themselves in
>> precise formal ?
>>> language... that is - after they learn LISP :) Aneil
>>
>> I don't believe this for a minute, I personally believe it is
>> possible to
>> develop a language that is biology friendly but at the same time
>> precise.
>> Jarnac is pretty simple, I know a lot of biologists who are happy to
>> use it
>> (and do things with it I never envisaged). Python is another example
>> of a
>> language that is simple to learn and use, I teach it to completely
>> newbie
>> biolgists whose only contact with a computer has been email and the
>> internet.
>>
>> I do believe there is a need for some kind of extensible non-xml text
>> based
>> langauge for systems biology, what it is I am not sure, but I'll be
>> interested to see b when it is released.
>>
>> Herbert Sauro
>
>
>
--
Marco Antoniotti http://bioinformatics.nyu.edu
NYU Courant Bioinformatics Group tel. +1 - 212 - 998 3488
715 Broadway 10th FL fax. +1 - 212 - 998 3484
New York, NY, 10003, U.S.A.
|
|
|
Posts: 961
Registered:
October 2003
|
|
|
Posts: 349
Registered:
September 2003
|
|
RE: Re: RE: Little b
|
29 Jul '05 10:23
|
|
|
Lisp is a bit of a jump for straight wet biologists. I've taught
computer programming to newbies for a few years now and it's difficult
to teach languages other than fairly straight forward one like basic,
python etc. Perl is quite a struggle for example.
H
-----Original Message-----
From: Marco Antoniotti [mailto:marcoxa@cs.nyu.edu]
Sent: Friday, July 29, 2005 8:28 AM
To: SBML Discussion List
Subject: Re: [sbml-discuss] Re: RE: Little b
Well, given that they learn to program in R (and Perl and Python and
Ruby and Tcl and SLDJ), they can also learn Lisp.
Cheers
--
Marco
On Jul 29, 2005, at 11:33 AM, Tomas Radivoyevitch wrote:
> Backing up what Herbert just said, biologists and MDs who took my
> stats class did indeed learn how to program in R.
>
>
> ----- Original Message ----- From: "Herbert Sauro"
> <Herbert_Sauro@kgi.edu>
> To: "'SBML Discussion List'" <sbml-discuss@caltech.edu>
> Sent: Wednesday, July 27, 2005 4:23 PM
> Subject: RE: [sbml-discuss] Re: RE: Little b
>
>
>>
>>> The "unnatural" structure of computer languages comes from the need
>>> for
>> precise
>>> unambiguous expression. If models are to become part of mainstream
>>> of
>> biology,
>>> perhaps it will be important for biologists to express themselves in
>> precise formal ?
>>> language... that is - after they learn LISP :) Aneil
>>
>> I don't believe this for a minute, I personally believe it is
>> possible to develop a language that is biology friendly but at the
>> same time precise.
>> Jarnac is pretty simple, I know a lot of biologists who are happy to
>> use it (and do things with it I never envisaged). Python is another
>> example of a language that is simple to learn and use, I teach it to
>> completely newbie biolgists whose only contact with a computer has
>> been email and the internet.
>>
>> I do believe there is a need for some kind of extensible non-xml text
>> based langauge for systems biology, what it is I am not sure, but
>> I'll be interested to see b when it is released.
>>
>> Herbert Sauro
>
>
>
--
Marco Antoniotti
http://bioinformatics.nyu.edu
NYU Courant Bioinformatics Group tel. +1 - 212 - 998 3488
715 Broadway 10th FL fax. +1 - 212 - 998 3484
New York, NY, 10003, U.S.A.
|
|
|
Posts: 349
Registered:
September 2003
|
|
RE: Re: RE: Little b
|
29 Jul '05 10:29
|
|
|
Unfortunately this was written after I saw the email from Mike, the
topic is now dead (as the preverbal parrot).
Herbert
-----Original Message-----
From: Herbert Sauro
Sent: Friday, July 29, 2005 9:24 AM
To: SBML Discussion List
Subject: RE: [sbml-discuss] Re: RE: Little b
Lisp is a bit of a jump for straight wet biologists. I've taught
computer programming to newbies for a few years now and it's difficult
to teach languages other than fairly straight forward one like basic,
python etc. Perl is quite a struggle for example.
H
-----Original Message-----
From: Marco Antoniotti [mailto:marcoxa@cs.nyu.edu]
Sent: Friday, July 29, 2005 8:28 AM
To: SBML Discussion List
Subject: Re: [sbml-discuss] Re: RE: Little b
Well, given that they learn to program in R (and Perl and Python and
Ruby and Tcl and SLDJ), they can also learn Lisp.
Cheers
--
Marco
On Jul 29, 2005, at 11:33 AM, Tomas Radivoyevitch wrote:
> Backing up what Herbert just said, biologists and MDs who took my
> stats class did indeed learn how to program in R.
>
>
> ----- Original Message ----- From: "Herbert Sauro"
> <Herbert_Sauro@kgi.edu>
> To: "'SBML Discussion List'" <sbml-discuss@caltech.edu>
> Sent: Wednesday, July 27, 2005 4:23 PM
> Subject: RE: [sbml-discuss] Re: RE: Little b
>
>
>>
>>> The "unnatural" structure of computer languages comes from the need
>>> for
>> precise
>>> unambiguous expression. If models are to become part of mainstream
>>> of
>> biology,
>>> perhaps it will be important for biologists to express themselves in
>> precise formal ?
>>> language... that is - after they learn LISP :) Aneil
>>
>> I don't believe this for a minute, I personally believe it is
>> possible to develop a language that is biology friendly but at the
>> same time precise.
>> Jarnac is pretty simple, I know a lot of biologists who are happy to
>> use it (and do things with it I never envisaged). Python is another
>> example of a language that is simple to learn and use, I teach it to
>> completely newbie biolgists whose only contact with a computer has
>> been email and the internet.
>>
>> I do believe there is a need for some kind of extensible non-xml text
>> based langauge for systems biology, what it is I am not sure, but
>> I'll be interested to see b when it is released.
>>
>> Herbert Sauro
>
>
>
--
Marco Antoniotti
http://bioinformatics.nyu.edu
NYU Courant Bioinformatics Group tel. +1 - 212 - 998 3488
715 Broadway 10th FL fax. +1 - 212 - 998 3484
New York, NY, 10003, U.S.A.
|
|
|
|
