Community Grids Lab Work

1
Community Grids Lab Work

• Marlon Pierce
• CICC Quarterly Meeting
• Jan 27, 2006

2
Some Notes on Our Approach

• Everything is based around Web Services.
• We build some.
• We inherit some
• UDDI, WS-Context
• We take some
• NIHs COMPARE and Cambridges WWMM suite.
• Services are fundamental units of action
• Clients to services can come in different
  varieties.
• Taverna workflow clients
• Portlet-based portal clients
• Other client environments and tools (like BPEL
  workflow engines) are compatible.
• All work with our same services.

Web Service
Web Service
Web Service
Taverna
Portal
3
Why Taverna?

• Taverna is an open-source workflow and annotation
  tool developed by the UK e-Science program.
• A nice client-side workbench tool for collecting
  services into a meta-application.
• Taverna works with generic Web Services
• WSIF helps you build client interfaces from WSDL.
• Designed to support bio-informatics applications.
• Also used extensively by our Cambridge
  collaborators.
• We anticipate extensive, ongoing support for it
  through the OMII.
• For example, maybe better Globus and OGSA-DAI
  client support.

4
Summary of Services

• Execution and File Web Services (J. Kim)
• Makebits, dvikm
• Supporting Wild Use Cases
• VOTable Services (S. Patel)
• Generic XML tabular data model from Virtual
  Observatory community.
• Potentially useful as intermediate data format
  for application (BCI, CDK codes) output and for
  screening center results.
• Sima will discuss a use case.
• Comes with parsers and a Java API.
• Wrapped now as a Web Service
• VOTables serve as data model.
• Table instances on remote servers can thus be
  manipulated.
• Inherited tools (VOPlot) can plot VOTable data.

5
Upcoming Work Globus and TeraGrid

• Globus services will greatly simplify running
  codes on high performance computing environments.
• Secure job submission, support for PBS, secure
  remote file management.
• Available on TeraGrid.
• Dvikmeans and some CDK codes should benefit.
• It has been a little problematic to develop
  Taverna clients.
• Not currently well documented
• We are working on intermediate Globus services
  for this.
• Wrapping the Java CoG Kit for protection from
  Globus version changes.
• And the CoGs nice API.
• But expect UK efforts in long term to replace.

6
Cambridge Tools

• Downloads put together by Nick Day
• http//wwmm.ch.cam.ac.uk/download/
• A bundled Tomcat server that includes all
  services and also portlets (i.e. service clients
  for browsers).
• Services listed on next slide.
• Portlets are compatible with OGCE, LEAD, and
  other work that IU has done.
• A version of Taverna that includes CDK and
  Jumbo51 tools.
• Some workflow examples (Jake will demo)
• We really would like to explore some applications
  using these but need help from chemists,
  cheminformatics folks.

7
Jumbo and Open Babel

• Jumbo is Peter Murray-Rusts suite of Java tools
  for manipulating the Chemistry Markup Language.
• Can also form the basis for CML Web Services
• Open Babel is a cross-platform program and
  library designed to interconvert between many
  file formats used in molecular modeling,
  computational chemistry, and many related areas.
  
• Wrap with Web Services

8
CDK the Chemistry Development Kit

• CDK is a Java-based, object-oriented set of
  chemistry libraries and packages.
• Model-view-controller design to separate
  structure from representation (i.e. 2d, 3d, text
  for same data model).
• Data classes element, isotope, atom, bond, etc.
• Allows you to programmatically build your
  molecules, etc.
• Lots of tools
• Format converters, rotators, translators, surface
  area calculators.
• CDK forms the basis for several other
  applications
• Jchempaint Java editor for 2D chemical
  structures.
• Seneca distributed/parallel application for
  computer-assisted structure elucidation (CASE)
• Nmrshiftdb open, online database for organic
  structures and their nuclear magnetic resonance
  (nmr) spectra.
• Can form basis of yet more services....

9
WWMM Web Services

• XindiceServer
• OBServer (i.e. OpenBabel)
• Mopac
• JumboToolServer
• JumboConverter
• INChiServer
• INChiGoogle
• Gamess
• Exist
• CMLRSSServer
• CML2SVG
• CML2Jmol
• CML2Gamess
• CDX2CML
• Build3d

10
Some notes from Peter Murray Rust

• InChI Generator - this will generate a unique
  identifier for a molecule. This will increasingly
  be the unique handle for retrieving molecules
• Openbabel will convert from any format to any
  other format
• InChIGoogle searches for any molecules on the
  web that have InChIs
• The value of any particular service will
  increase proportionally to the total number. A
  chemoinformatics app consists of tens of
  different functionalities. Other things in our
  pipeline are 2D layout, 2D image generation, etc.
  By the time all these add up then there is quite
  a lot! --PMR

11
Web Site, Blog, and Mailing List

• The web site, www.chembiogrid.org, is plain old
  HTML (for simplicity and long-term
  maintainability) backed by a WebDAV server.
• You can use DAV clients (built into Windows, for
  example) to log into the server and upload web
  material.
• Our DAV server runs inside of Tomcat, so it can
  serve up HTML, JSP, and similar stuff.
• But not Perl, PHP, CGI or similar.
• Its virtually hosted through Apache, so if you
  really want PHP, we can work something out.
• We also set up a Wiki....
• We are blogging related stuff (on VOTables,
  Taverna Life Lessons, etc) at http//communitygrid
  s.blogspot.com.
• I think this is a useful way to post notes.
• We can create a specific blog for this project if
  you want.
• Mailing list cicc-dev-l_at_listserv.indiana.edu.
• Contact me if you want to subscribe.