[Pipet Devel] Re: loci

[J.W. Bizzaro]

Harry Mangalam wrote:
> I'm considering releasing the next version under std GPL, but I'm old enough
> to want to take a good look at it and try to consider the possibilities that
> GPL requires/allows.

Okay.  I hope you do!  I started out worrying about what greedy corporate
types might do with my programs, but I don't care as much now.  I think the
GPL gives pretty good protection to the intellectual property of the
developers.  And I think an important part of an unestablished project is
_not_ restricting one's work to anyone.

> 
> Right - it's a vanilla ANSI-C command-line app - one of the reasons for
> dev'ing it was to supply something like a DNA Strider for the command-line.
> I develop on Linux and port to IRIX/SunOS/Solaris/DEC Unix/HPUX with no
> problems.. yet.

That's fine.  I think ANSI-C is more portable than C++ and is more suited to a
UNIX environment.  We use ANSI-C to supplement Python.  But Python is
preferred here because of some very powerful features.  It is also, from my
observation, the scripting language most preferred by physical scientist (I
sound like an advertisement)...considered by a few to be a likely replacement
for FORTRAN.

> 
> I was considering starting with the former (a GUI-wrapped command-line app)
> and moving to the latter (fully GUI), but I'm still feeling my way in terms
> of how to present it.  As I understand LOCI, the underlying apps can be
> distributed but communicate at the XML layer.  I'm starting to add this to
> tacg for reasons related to interoperability but until yesterday, unrelated
> to LOCI :).

Yes.  Tools can/will communicate locally via (1) direct Python implementation
or (2) indirect use of XML.  The other way to communicate is (3) remotely via
XML with a CGI-like interface.

> Are you planning to make a psuedo-visual programming language out of this -
> is this what the GCL is?  If I understand XML correctly, this shoul dbe
> possible but would probably require a large expenditure of energy...

Yes.  I was just talking to Carlos Maltzahn in our group about that (Carlos
developed the Paos Project for distributing Python objects over a network, and
he will help incorporate that into Loci).  GCL is about the highest level
programming language you may find, and it is specifically to manage multi-step
analyses in a graphical way. (I don't consider biologists to be very
computer-savvy.)  The XML is mostly used to format data and not to
issue/manage commands.  The job of GCL is to issue/manage commands.  But
putting commands in with the XML is an interesting thought.

> 
> I like the idea of being able to use whatever underlying language you want -
> lots of goodies are written in perl and there's no real reason to exclude
> those apps/libs and those authors.

Yes!  The distributed model with a CGI-like mode of operation will make this
connection between Loci and any other command-line language!


> I also am working on contract for NCGR (National Center for Genome Resources
> (Santa Fe) which is also interested in developing freely available tools for
> sequence analysis/bioinformatics and I'll try to get them to pay attention to what you're
> trying to do - there may be room for some effort on their part.

What sort of "effort" are you speaking of?  Loci is unfunded, but that may change.

> I tend to agree with your feeling that both can go forward - with the state
> of funding these days, there will be times when one or the other will be
> moving faster, but both groups should stay in contact with each other.  It
> would be good if representatives of both could meet and get drunk together
> at some birds-of-a-feather meeting soon...

:-)  Which continent?

> OK - I am VERY much interested in hearing what the rest of you think about
> how this should be done - I'm interested in getting biosequence analysis
> made much easier and cheaper and have tried to walk my talk by taking the
> time to put something together towards that end.  If I can contribute to
> other projects by this, so much the better.

IF you go GPL, I would hope that tacg would be a model for taking a
command-line bioinformatics program and adding a GUI to it.  In other words, a
good point of collaboration for us would be to use the energy that would be
put into making a GTK interface for tacg, and put it into developing the part
of Loci that tacg would need.  We would otherwise be duplicating our efforts. 
We are just starting to make these tools, such as sequence visualization and
editing tools, so why not make them to work for tacg (as well as some EMBOSS
programs)?  Thomas Sicheritz (author of BioWish) is working on an editor right
now, and I think Justin Bradford may help with XML implementation.

> 
> So by all means, plase keep me in the loop.  If I can help out in any way.
> let me know...

Okay.  You are hereby on the mailing list.  And I'll consider you an observer
who may want to help (please help :-) with incorporating his not-yet-GPL
analysis algorithms.



Jeff
-- 
J.W. Bizzaro                  Phone: 617-552-3905
Boston College                mailto:bizzaro at bc.edu
Department of Chemistry       http://www.uml.edu/Dept/Chem/Bizzaro/
--