David Wild Update June 2006

From Chemical Informatics and Cyberinfrastructure Collaboratory

Contents 1 Web Service Infrastructure 2 Workflows 3 Data Mining of DTP Database 4 Visualization and Interaction Tools 5 Methods 6 Outreach 7 Six Month Outlook 8 Points for discussion

Web Service Infrastructure

• New services
  • DTP Database Screen Search
  • Protein Database
  • ToxTree
  • New VOTABLES services (TAB->VOTABLE, VOTABLE->TAB)
  • Distributed Drug Discovery in progress
• Decision to use tab files (for the moment)
  • Mk.1 Tab File, Mk.2 VOTABLES, Mk.3 CML?

Workflows

• Implemented key HTS Data Analysis workflow (Workflow 2):

SCREEN SEARCH -> FILTER -> (TOXTREE) -> DIVKMEANS -> VOTABLES -> VOPLOT

• Taverna 1.4 now available including visual workflow builder
• Pilot project with Plale/Gannon to illustrate how our Taverna framework can be easily integrated with other environments
• See Workflows for more information

Data Mining of DTP Database

• Major review of current work, and identification of areas of opportunity
• Characterization of database in terms of diversity, compound profiles, similarity to HTS datasets
• Collaboration with Faming Zhang

Visualization and Interaction Tools

• First portlet application in Gridsphere - BCI clustering portlet
• Developments to PubChemSR - similarity searching
• Usability experiment on Pubchem/Chmoogle/ChemDB completed. Developing wider package of usability experiments (outside grant...)

Methods

• Demonstrated ability to cluster PubChem in 5-6 hours on AVIDD (20 procs)

Outreach

• Begun Collaborating with Michigan MACE
  • Wrapping around PDBBind Dataset
  • Mookie - Heather Carlson
  • Late Summer Collaboration Meeting
  • See Indiana Michigan Collaboration for more information
• Potential connection with OpenEye
• Potential connection with Jake Chen
• ACS presentation accepted for September
• Microsoft presentation September/October
• Publication (Wild, Wiggins) in Drug Discovery Today, May 2006, on chemoinformatics education (in addition to JCIM article)
• Publicaton (Wild, Wu, Zaharevitz) on DTP Data Mining, submit July
• Publication on Web Service Infrastructure, submit July

Six Month Outlook

• Summary so far
  • Implemented 12 chemoinformatics web services (each service can provide multiple functions)
  • Implemented 2 drug discovery related workflows using these services
  • Several prototype visualization / interaction tools
  • Started mining the DTP Tumor Cell Line Set
  • Demonstrated feasibility of clustering entire PubChem dataset
  • Begun collaborations with PMR group, Michigan MACE, Faming Zhang, DTP
  • Posters at spring ACS, presentations at fall ACS and Microsoft, 2 papers in preparation

• Must have
  • Scientific successes using workflows
    • Faming Zhang Kinase collaboration
    • DTP data mining with PDB
    • Analysis of MLSCN data in PubChem (Workflow 2)
  • Expanded Workflow 2
    • Add in 2D structure viewer to VOPlot
    • ToxTree
    • Any QSAR models available
    • PDBBind
  • 2 more workflows
  • A non-trivial portlet interface talking multiple workflow streams
  • OSCAR success with PMR collaboration (Matt Stahl concurs)
  • Publications on DTP Data Mining and Web Service Infrastructure
  • Some success with Michigan collaboration (PDBBind, Workshop, Chemoinformatics Course)

• Nice to have
  • Execution environment for Taverna, including ability to wrap workflows as web services and possibly BEPL support
  • R statistical and QSAR services
  • Something with a .NET interface
  • Scientific Workflow using reaction database from Distributed Drug Discovery
  • Publication on DivKmeans

Points for discussion

• Use of OEChem and potential for NIH buyout
• Joint workshop with MACE