myGrid Steering Committee Meeting

1
myGrid Steering Committee Meeting

• myGrid Team Leaders
• Sheffield University
• 16 January 2004

2
WP1 Bioscience support

• Peter Rice

3
Bioscience support current status

• Life Sciences Identifier (LSID)
• Revised submission to OMG
• lsr/03-12-02
• Bioinformatics workflow support
• Working with WP-5
• X-Scufl
• Taverna
• Freefluo

4
Bioscience support plans

• LSID
• Drive OMG process to adoption
• Expose more services
• Soaplab
• Data services
• Third-party services
• PathPort, BioMoby,
• Discovery via registry

5
WP2 - Architecture

• Luc Moreau

6
WP2 (status)

• View Service
• UDDI functionality
• RDF based service descriptions
• Protocol to add metadata
• Registration of semantic descriptions of both
  services and workflows
• Notification of change
• Notification service
• Proper publish subscribe model
• Persistent topics, semantic topic,
• Security
• First questionnaire published, results collated
• Second questionnaire (about) to be distributed

7
WP2 (next 6 months)

• View Service
• Personalisation (views, curation, notification,
  policy)
• Federation
• Using functionality
• Notification service
• Semantic topics
• Federation
• Customisation (policy based)
• OGSA
• Security
• Distribution of second questionnaire and result
  collation

8
WP2 (end game)

• For us, shortly after 6 months
• Robustness bring some elements of access
  control
• New opportunities
• OMII
• Best practice
• PASOA more focus on provenance

9
WP3 Data Management

• Paul Watson

10
WP3 Strands

• Bioinformatics Scenarios
• Information Repository
• Distributed Query Processing
• these are discussed in turn.

11
Bioinformatics Scenarios

• Past
• Graves Disease Scenario workflows designed and
  implemented
• Recent
• five papers written
• constructed workflows with 3rd party services
• collaboration on microarrays with Yale HK
• Future (March)
• incorporate information extraction, DQP
• microarray analysis services
• Future (End of Project)
• add notification semantics
• apply ideas to new scenario

12
Information Repository

• Past
• built old IR still in use
• Recent
• designing new IR based on Information Model
• document out for comment
• Future (Mar)
• first version of new IR
• Future (Sept)
• enhanced Version
• Future (by end of project), options are
• security, mgt queries, data mining, notification

13
Distributed Query Processing

• Past
• public release in September 2003
• used in Graves Disease Scenario
• Recent
• investigation into better integration with MyGrid
  
• Future (March) Generic Wrapper
• Future (Sept) tooling to generate WS wrappers

14
Summary

• Work on Information Repository is high priority
• Less DQP work from now on
• focus moves to ease of use
• but will continue in DAIT project
• Graves Disease Scenario continues to drive
  bioinformatics

15
WP4 Metadata management

• Carole Goble

16
WP4 portfolio

• Ontologies to publish and discover services and
  workflows.
• Ontologies to guide workflow composition.
• Ontologies to annotate, discover and link any
  myGrid object.
• Models of metadata for myGrid, and their
  deployment using semantic web technologies.
• Provenance models, architecture, implementation
  and browsing.
• Williams Syndrome development and support.
• Phil Lord, Chris Wroe and Mark Greenwood at
  Manchester
• Four PhD students
• Close relationship with Southampton team

17
Semantic finding

• Progress
• Service and Workflow semantic discovery
• Software rewrite and consolidation, Testing and
  documentation
• Service and workflow publication preliminary
  software based on PEDRO
• Ontology updates for new services
• 3 papers, I3C and MOBY-DIC Hackathons, grant
  proposal
• Short Term
• Complete migration to OWL from DAMLOIL for
  software
• Simple single ontology deployment via static RDF
• User tools service describers, lexicons
• LSID integration
• Ontology maintenance
• Semantic descriptions of topics for event
  notification.
• Medium Term
• Full OWL deployment using OWL server and instance
  store
• Multiple ontologies strategy
• User expansion of ontology
• Endgame
• Discovery and annotation of any object in myGrid

18
Info model and Provenance

• Progress
• Designing integrated conceptual model
• Supporting exported provenance from services,
  integration framework, layers of provenance
  models execution, knowledge.
• Piloting design with Williams syndrome case
  study
• Presentation, querying browsing with RDF and
  Haystack
• Integrated LSID resolver launched from provenance
  examination
• Experiments with linking execution logs using
  ontology
• 2 publications, PASOA associated project
• Short-medium term
• Assist in deployment of metadata model, in
  particular RDF and links to ontology and LSID
• Provenance presentation -- Haystack
• Templates for outcome provenance generation
• Endgame
• Full demo and deployed model.

19
Williams Syndrome

• Progress
• Williams Syndrome workflows and provenance
  browsing
• ISMB paper
• Short-Medium Term
• Continue support
• Batch execution and iteration
• Provenance generation
• Provenance presentation and integration
• Event Notification
• Bringing in new services
• Using the mIR and workbench?
• Population of view service with workflows,
  services and personal views.

20
WP5 - Workflow

• Matthew Addis

21
Progress over the last 6 months

• Freefluo and Taverna functionality
• Looping, conditionals and nested workflows
• Data model using collections and MIME types
• Easy to do workflows involving data sets
• Enhanced result and provenance browsing
• Dissemination and uptake
• Done Link-up, AHM, GGF LSWG, I3C hackathon, NeSC
  workshop, support to Newcastle and Manchester,
  Taverna user and developer groups, direct contact
  with users
• In the pipeline WWW, Bioinformatics journal, ISMB

22
Current status

• Taverna and Freefluo are now well established
  open source projects
• Running for about 9 months
• Active user and developers groups
• Academic and commercial interest/uptake
• Development is driven by user requirements
• More than enough to do for the rest of myGrid ?

23
Current users and developers
24
Plans for the rest of the project

• Development
• Improved robustness and error handling
• Long running workflows
• Integration of a wider range of apps and services
• New provenance model
• Integrate semantic service discovery
• Dissemination and uptake
• Direct interaction with external groups/ and
  projects
• Papers, conferences and workshops
• Newsgroups and word-of-mouth
• Collaboration and further work
• Sisters project, OMII
• Best practice project?

25
WP6 e-Science Layer

• Chris Greenhalgh

26
E-Science Layer Achieved

• Since previous meeting (6 months)
• Bug fixes, stability, All Hands demo
• Improved UserProxy and notification client
  configurability
• Release 1.0
• Separated Gateway module from Workbench
• Prototype LSID interface to mIR v.3
• Started to build links with local (Nottingham)
  biologists/bioinformaticians

27
E-Science Layer Current State

• Existing Workbench (NetBeans-based)
• Some issues moving to latest Taverna releases
• Mina bogged down in teaching (notification
  client, collaboration requirements)
• Emerging problems with NetBeans flexibility and
  paradigm
• Prototype explorations
• Haystack as LSID/RDF interface
• Structural (text) record parsing
• Knowledge search and extraction from records
• Simple interactive graphs
• Further exploration of framework/support user
  requirements
• Integration, presentation, interaction

28
E-Science Layer Options

• Interface technology
• NetBeans/Workbench as is
• Or major re-engineering
• New graphical interface (e.g. 23D)
• (New) Web portal
• Incidental interface (e.g. Web interceptor c.f.
  COHSE)
• Collaboration
• Via mIR/notifications
• Synchronous communication support
• Single display groupware
• Interaction (e.g. Gateway)
• Domain type and collection support

29
E-Science Layer Future

• EPSRC Pilots (March 04)
• Support new mIR interface/framework
• Support (parts of) new Information Model
• All Hands (Sept. 04)
• Increased interaction framework support
• One new interface option
• Incremental evaluation with local users
• Closure (March 05)
• Fluid malleable interaction prototypes
• Cross-links to other eScience activities

30
E-Science Layer Other

• Dependencies
• New information model
• New mIR design/implementation
• Enactor, Notification,
• Risks to success
• Time required for tracking changes, bug fixes and
  minor enhancements vs. new capabilities
  research
• Lack of clarity
• Publications
• Mainly supporting other publications
• Downstream Fluid interaction? Collaboration?
  Contextual support? Semantic-driven interface
  issues?

31
WP7 Text Services

• Rob Gaizauskas

32
Text Services Aims Objectives

• Aim to provide novel text access capabilities to
  biological science researchers as a service
  within the myGrid platform
• Data for a variety of biological scenarios of
  interest will be mined/extracted from text
  collections
• Entities (e.g. genes, proteins, residues,
  species)
• Attributes (e.g. residue function)
• Relations (e.g. locations, interactions,
  pathways)
• Extracted data and source texts will be
  accessible
• Via the MyGrid workbench/UI
• Via web services
• Enhanced data presentation mechanisms will be
  explored, such as
• Context aware delivery (ambient text)
• External linkages
• Structuring, clustering and internal linking
• Notification services

33
Recent History

• Benchmarked AMBIT implementations against PASTA
• Completed Termino
• Very large scale, flexible terminological
  database
• Mechanism for compiling out Finite State
  Recognisers
• Generic tool to populate Termino from external
  sources (e.g. UMLS)
• Design implementation of full-text search
  engine (MedCow)
• Simple boolean search engine for large
  semi-structured text collections
• Supports field-restricted search
• Two interfaces Browser GUI SOAP/web-service
• Design implementation of a database to hold
  Medline.
• Design implementation of AMBIT Query Language
• Supports complex search over multiple,
  heterogenous text resources

34
The State Of Play

• Indexing of structured Medline data for database
  search and access
• Indexing of last 10 years of Medline for MedCow
• SOAP-based access to AQL engine
• AQL integration of MESH MedCow search of
  Medline
• AMBIT scaled to handle arbitrarily large text
  collections
• Participation in BioCreative challenge

35
Next Steps

• March
• Medcow indexing of the complete Medline database
• AMBIT processing of demonstrator texts to extract
  selected entity types and relations
• Browser-based interface to AMBIT
• Further integration of AMBIT into myGrid
  Demonstrators
• September
• AMBIT processing of all of Medline to extract
  selected entity types (proteins, genes, etc)
• Term induction and clustering
• Further Integration of AMBIT into myGrid
  Demonstrators
• Endgame
• AMBIT processing of all of Medline to extract
  selected entity types and relation types
• Extend entity and relation type coverage
• Further Integration of AMBIT into myGrid
  Demonstrators