[DAS2] Notes from the weekly DAS/2 teleconference, 6 Nov 2006
Steve_Chervitz at affymetrix.com
Mon Nov 6 19:16:49 UTC 2006
Notes from the weekly DAS/2 teleconference, 6 Nov 2006
$Id: das2-teleconf-2006-11-06.txt,v 1.1 2006/11/06 19:13:26 sac Exp $
Note taker: Steve Chervitz
Affy: Steve Chervitz, Gregg Helt, Ed Erwin
CHSL: Lincoln Stein
Dalke Scientific: Andrew Dalke
UAB: Ann Loraine
UCLA: Allen Day, Brian O'connor
Action items are flagged with '[A]'.
These notes are checked into the biodas.org CVS repository at
das/das2/notes/2006. Instructions on how to access this
repository are at http://biodas.org
The note taker aims for completeness and accuracy, but these goals are
not always achievable, given the desire to get the notes out with a
rapid turnaround. So don't consider these notes as complete minutes
from the meeting, but rather abbreviated, summarized versions of what
was discussed. There may be errors of commission and omission.
Participants are welcome to post comments and/or corrections to these
as they see fit.
* 2.0 spec freeze discussion
ls: hapmap project repercussion: instability of das/2 spec has lead
for me to recommend against using it for hapmap db. Have had to put
hapmap data onto a soap server. Need to have an internal project,
would have control over the process. not best possible protocol, but I
could promise delivery of working s'ware on a reliable schedule.
ls: I had a real deadline to deliver an adaptor to the NCI by end of
Nov, w/o having a spec that is in stone that I can write to, can't
deliver by that date, and can't get extension. went on record a year
ago saying get spec was stable, good to build on, and it's not. would
like to ask that we freeze the spec, remain frozen, the next version
be das/3 and we guarantee das/2 is frozen for at least 2 years.
gh: ok with that. how do other's feel.
ls: if brian gilman can write a das/2 adaptor for cabig by end of nov
based on spec now, it's not a crisis. we have two dependent things:
(1) is a das/2 adaptor for caCORE that can read das/2 sources,
(2) das/2 server for hapmap and das/2 server for vertebrate promoter
db. NCI will not accept delivery of das/2 adaptor after end of Nov. If
so, then the other two projects (servers) would become irrelevant, and I
would withdraw from those two as well. NCI decided the spec was never
going to stabilize, so wasn't flexible in giving more time past
ad: brian wants schema in XSD not rnc. some changes: source -> method
ls: two things: (1) is spec changing too much? this conversation was
to create enclosing tag to create a group of related feats for
streaming purposes. (2) perception issue: killing me because the NCI
people read the archives and can see that the das/2 spec is in thrash
and is not converging.
ad: parent field, adding one single attribute to each feature, not a
major alteration. Then we have discussion in ensuing two weeks
following teleconf. we can freeze rnc schema now, everything works now.
aday: html document is out of date, not sure what's in schema.
ad: I sent it around a while back.
aday: haven't seen it.
gh: I read it. easier to read it than the html doc.
ad: html doc doesn't get touched because it's much harder to write.
ls: then freeze the rnc, remove html, point people to RNC.
gh: nothing wrong with html, but it should say the formal spec is in
ad: rnc it's not complete (e.g., reference genomes are defined on a web
page someplace). schema is not going to be the spec. there are
somethings that schema definitions can't describe.
gh: should have a pointer to the rnc at top of html and say "it is
frozen and will stay frozen as das 2.0"
[A] place DAS 2.0 frozen notice on html spec doc (after 1-2 day analysis)
gh: salvagable situation with NCI?
[A] lincoln will notify NCI of DAS/2 schema freeze
ad: before people can say yes, i need to check in unified version of
the schema, then folks can sign off on a unified document. current doc
is in 8 parts. i'll put the unified schema into version control.
ls: it's in the 'draft3' subdir.
[A] Andrew will consolidate rnc schema document, check into cvs, notify list
aday: describes formatting of xml and what each fields do. incomplete
to impl a server because it doesn't describe req/response cycle.
ad: yes, this is a description of the format.
aday: it's an incomplete format.
ad: describes the stuff needed to be returned back from the server, so
it's complete from a server implementer's point of view.
ad: what is in the html that doesn't agree with schema?
aday: property response, fasta. timestamp on the html doc has been
changed 10/24/06. Need to read this again.
ad: I made some minor changes while at the EBI.
gh: freeze the schema, freeze the intent of the html to smooth out
clarifications, all devs read both schema and html, OK with freezing
it in the next day.
[A] freeze schema as DAS 2.0 (get), freeze html intent and clarify, by 7 Nov
gh: in light of that, improving the biodas site. andreas suggested
turning into wiki doc. need to allow multiple people to edit.
sc: I tend to do most biodas.org upkeep. bioperl has migrated to
a wiki format. can probably borrow their template and set up something
similar for biodas.org.
[A] steve will convert das site into wiki style site
ad: typos and xml mistakes in april to the interaction document
(writeback) in the last 6 mos.
gh: not talking about freezing the writeback portion.
ee: is ucla das/2 server working now, top level doc, can't use it via
aday: brian is testing against the affy server, that sources doc is
still not responding.
[A] allen will fix sources doc on ucla server
ad: proxy work (email) accept das/1, interface with das/2. Serving
das/2 from a das/1 server. initial result was slow with python's
templating lang. new stream based parser with stream based output for
doing it. in progress now.
gh: what about auto testing of das/2 servers from his registry. talk
to andreas about it? ping for alive-ness?
ad: should still work easily, can't remember what andreas said about
[A] gregg will ask andreas about live-ness testing das/2 servers via
gh: uri's that affy server returns only work for a single version of
each genome (latest version). trouble with xml:base that was partially
fixed last week.
[A] gregg/steve will fix affy server xml:base to support all genome versions
gh: stabilize spec, read and sign off the spec, need to address and
stabilize. when funding agencies start pulling plugs based on das/2,
this is serious.
ls: these are management consultants. if promised s'ware product cannot
be delivered in working order in time expected, so they make the calc
that it's better to cut their losses.
ls: need a human readable html doc that's consistent with the rnc
document. public declaration on the website that people can rely on it
for 2 years.
bo: Any developer will want to see this, as well as a reference
gh: read html doc today/tomorrow, with an eye towards agreement with
schema. I don't think it's that far off.
[A] everyone read html doc for agreement with schema, finish by 7 Nov 2006
Other topic: DAS-related projects
al: NSF plant science cyber infrastructure project:
ls: univ of georgia, malmberg. Another one I'm doing in collab with
al: incorporating anything from das, or viz work?
ls: univ of GA plan, my role is in annotating plant pathways.
in my project: all community annotation using wiki, kind of a
plantopedia. text is not very structured, series of pages which have
some constrained fields, genome annotation, genotypes, ontologies,
everything else is text annot on top of it. reason: natural language
has gotten good. people should not start dumbing down their
communication with computers, but communicate in english.
al: dense and compact abstract text.
ls: people are identifying regions of text in an xmly way.
al: proposals for centers, akin to ncbi for plant biology... is the
plant wiki idea to be a component?
ls: three main parts....
al: do people on das want to write something up for the cyber
infrastructure? das/2 seems appropriate.
ls: should talk about das being part of it. proposing an open source
api, basically a bus, that allows you connect the consumers of data
with producers of data. a s'ware layer that goes over an opaque
transfer protocol. deliberately not be cross platform. A s'ware kit on
laptops, prepopulated, autoupdates, spec that s'ware devs can write
to. in terms of an xml protocol that people can plug into, people are
never going to see that layer.
al: like an OS.
ls: in fact called the plant OS.
al: terrified that nsf will give 1-2 groups all the funding and we'll
have a monolithic structure. would like to try many ideas, let free
market decide. How can we give the people who do the hard work enough
funding to keep them involved, esp if they have 40hr/wk jobs as well.
ls: they don't want cyber infrastruct proj to make awards to people.
that would be taking over NSF's role.
ls: only seven page write ups.
al: I want to propose viz using das and microarray data. webservices and
microarray data. takers?
aday: webservices yes.
gh: other commitments now.
bo: mark carlson has been integrating das client into MeV.
al: incorporate into lincoln's proposals?
ls: problem now: i've identified who the PIs are on the
project. Microarray viz=owen white at tigr. can't change it now.
ls: An even bigger project that TIGR is involved in now is the
biofuels initiative - $250M over 5 years. very important project,
bigger than cyber infrastructure proj. biologist, engineers, nanotech,
al: do we want better das servers or cheap fuels?
ls: not for cheap fuels, but global warming.
gh: to wrap up this meeting, html documents can be edited out of the
repository, latest get specs and rnc docs. no need for an editable web
ls: andrew is going to freeze the rnc, then make one pass over the
html, then open it up to all devs? could be chaotic via source code
gh: will confer with andrew over the plan and let folks know.
[A] Gregg will inform all how to proceed re: html doc editing.
[A] Next DAS/2 conf call next Monday (13 Nov, 9:30am).
More information about the DAS2