[DAS2] DAS/2 weekly meeting notes from 5 Dec 2005

Steve Chervitz Steve_Chervitz at affymetrix.com
Mon Dec 5 20:42:44 UTC 2005

Notes from the weekly DAS/2 teleconference, 5 Dec 2005.

$Id: das2-teleconf-2005-12-05.txt,v 1.2 2005/12/05 19:55:54 sac Exp $

Note taker: Steve Chervitz

  Affy: Steve Chervitz, Ed E., Gregg Helt
  CSHL: Lincoln Stein
  UC Berkeley: Suzi Lewis
  Sanger: Andreas Prlic
  U Alabama: Ann Loraine
Action items are flagged with '[A]'.

These notes are checked into the biodas.org CVS repository at
das/das2/notes/2005. 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. 

Today's topic: Implementation status reports

* Planning new IGB release in 2-3 weeks
* Revised IGB docs
* Updating the IGB DAS client
  - Want to make the DAS/1 iface more like DAS/2. Currently, they are
    isolated in the code and the interface. Maybe done w/in next 2 weeks.

Question: Has anyone use the agile 2D java stuff?
( http://www.cs.umd.edu/hcil/agile2d/ )
AL: Hasn't used it but used jazz (piccolo). Couldn't figure how to get it
to zoom in only 1D. they do 2D.

GH: Looking to speed up rendering in IGB. The Affy transcriptarium loses
hardware acceleration when using normal java 2d/swing stuff (one cpu
controling 8 monitors).

agile 2d background: Billed as a drop in replacement for java 2d that runs
on top of open GL. Significant speed ups on solaris, other
platforms. But Gregg finds it is slower on windows. Requies open GL
for java that does lots of native stuff underneath. May work on
mac OS X.

The focus on windows b/c in order to use spotfire and other
windows-only apps. 

AP: Does it use double buffering?
GH: Yes. Swing does this. Builds volatile image. Hooks into hardware
acceleration graphics card. If you can construct a volatile img,
rendering goes through graphics hardware, which is a big speed up.
But swing can't deal w/ multiple graphics cards (Direct X limitation?).

Open gl may deal with things better in this regard.

AL: Maybe get a consultant. This is very specialized stuff.

GH: Getting some time from sun java graphics engineers could help.

Ed E
* Debugging for DAS/1 in IGB client
  - Fixed bug in the Affy DAS/1 server.
  - Also fixed a few bugs in parser (concerned note, length tags -- not
    heavily used). 

* Working on hooking up tomcat to apache to enable apache to handle
  requests that are fed to an underlying servlet engine.
  - Have made progress but need to do more testing before installing
    on netaffxdas server machine.

GH: Ultimate goal is to release the Affy DAS/2 server as a servlet
    using standard configuration. Isn't this sevlet-under-apache
    configuration fairly standard?

AP: We provide a war file. users can plug this into their app server.

SC: The apache-tomcat config is quite common for ISPs or other
situations where you need to redirect apache to different servlets
depending on different companies, products, etc.

SC: Assuming the apache configuration is ready, is the DAS/2 server
code ready to go? 
GH: depends on how much of the spec we want to support. Estimates 6mos
to 1yr before releasing our server. Also probably just release DAS/2.
DAS/1 is also just partial and does some custom things. Better to
focus on fleshing out DAS/2 server.

SC: Possible complexities for apache/tomcat due to our port forwarding
(DAS/1 and DAS/2 servers running on different ports).
GH: There's no reason we can't run these servers on different

GH: Steve's time for DAS work?
SC: Have to transition back to NetAffx now, but can still proceed with
DAS work, just less cpu cycles. Can still take and post notes.

* Working on new arabidopsis 2010 proposal from last year.
  - Main aim of grant: setting up data archive for arab, and for
    people to add to it. Currently scattered. people put data on their
    website and move on. No easy way to collect it in one place.
  - Setting up a web service at UAB, give it an ID, info about id
    (e.g., whose affy, tair, agi). The server will return synonyms for
    that ID. Goal is to be a backend for IGB to do searches for an
    ID. Will talk to the UAB server via some data format.
  - Grant will include funding for uab and Affy (usability
    enhancements for IGB as front end for data repository, figuring
    out protocol for talking to id server)

GH: Will you run DAS/2 servers at UAB?
AL: Yes, we'll set up the servers, but not do server development
itself. Can test and offer feedback from installation issues, etc.

GH: Will data repository be in Alabama?
AL: Yes. We have good sysadmin support. Originally TAIR would do it,
but we've built things up since then. Grant will be more smaller scale
than last year. No comparative genomics viewers. Sticking to core
expertise. Focus on getting one genome right.  Also funding for
genoviz sdk. library upkeep. tutorials.

GH: Reimplement IGB on top of picolo!
AL: Looking to Affy on fancy graphics stuff since they know the
libraries better than anyone else.

GH: Regarding your synonym resolver - have you talked to suzi about
what they're doing?
AL: No. Looked at the flynome link steve mentioned
(http://flynome.com). If there's a standard out there we'd love to use

LS: There is Gene Seer developed here - a synonlym database for
gene IDs. Supports many to may associations between synonly and the
canonical forms. No interface besides a web-form. May make sense to
provide access to it via DAS/2.
GH: layering a DAS/2 interface on top of the synomym service might be

[A] Ann will help develop gene synonym server on top of DAS/2

EE: In IGB we have a need for synomym service for genome builds
(e.g. hg16 = build34). Currently a hack in IGB, would be nice to do
via DAS.

AL: Plan to start with arab first.
GH: Using Gene Seer scheme for your storage you will get a lot of other
genomes for free.
AL: Is it open source?
LS: Yes, but it's not my code (ravi at cshl.edu) heavily used in cshl. Loaded
with all model orgs. This is now public: http://geneseer.cshl.edu
- published in bioinformatics.
- no arabidopsis (rat, mouse, human, celegans, yeast).

[A] Ann will contact Ravi re: Gene Seer & get letter to submit w/ her

SC: Does Gene Seer contain LSIDs?
LS: No, but could be added as a synonym.
Every name has a species, and unique type (embl, ncbi).
E.g., you can search for rad5 - in all species, or in yeast, or as a
locus name in yeast. Used for the RNAi libraries here.

AL: Is anyone else building such a resource?
LS: Google maybe....

* Not much coding for DAS/2. Waiting for issues with spec to be
  discussed. (next week). DAS/2 interface should be easy to add.

GH: Spec is changing, so impl is hard to do.
AP: Database and java server-side code has been running a while. A new
interface should be easy to do.

Andrew (via Gregg)
* Planning to get the registry-related stuff formalized in the spec soon.
* Working on setting up web service for validation on open-bio
server with Chris D.
* Updating spec but is slow going. Will join the call next week.

LS: Whose in charge of updating the spec now -- me, Steve, Andrew?

[A] Andrew is now responsible for updating the spec.

* Update on the NCI DAS/2 project:
  - Awarded grant for (funded by NCI) after 1 year negotiating.
  - caBIG grid depends on caBIO java lib/API from NCI. They had
    dropped all DAS support when going from caCORE v2 -> v3.
  - Brian Gilman proposes to add DAS/2 support to caCORE v3.
    o First, create plugin API for caCORE. Will allow you to add in
      class libs without recompiling java code. Permits caCORE to
      speak arbitrary protocols (~3 mos work).
    o Then implement DAS/2 plugin (~3 mos work).
    o Starting on 12/17.
    o Will use Alan Day's biopackages DAS/2 server (gmod), serving hap
      map through that.  Client will be caBIO.


[A] Discuss spec issues for next weeks teleconf: regions, seq, registry,

More information about the DAS2 mailing list