BiBiServ: Practices and Problems WebServices and Workflows on the - PowerPoint PPT Presentation

BiBiServ: Practices and Problems WebServices and Workflows on the Bielefeld University Bioinformatics Server Sven Hartmeier NETTAB 2007 - 13.6.2007 - Pisa, Italy Faculty of Technology - Bielefeld University

BiBiServ-Basics... • Bi elefeld University Bi oinformatics Serv er • founded in 1996 • hosts various BioInformatics Tools developed within several research groups at Bielefeld University

BiBiServ-Tools • currently around 30 tools in the areas of: • RNA Structure Research • Alignment Computation • Genome Comparison • Primer Design • Evolutionary Relationships • Miscellaneous Other Applications

Used Technologies • 1996: HTML/CGI • 2003: WebServices using XML-RPC • 2004: WebServices using SOAP / WSDL HTML Interface transformed to „just another WebService Client“

Architecture SGE Computation Grid WebService WebServer Server „BiBiServ“ „BiBiWSServ“ Database Server

Usage (HTML Interface) Request SGE Computation Grid WebService WebServer Server „BiBiServ“ „BiBiWSServ“ Database Server

Usage (HTML Interface) SGE Computation Grid WebService WebServer Server request „BiBiServ“ „BiBiWSServ“ Database Server

Usage (HTML Interface) SGE Computation Grid WebService response: WebServer Server Job-ID „BiBiServ“ „BiBiWSServ“ Database Server

Usage (HTML Interface) SGE Computation Grid polling WebService WebServer Server „BiBiServ“ „BiBiWSServ“ Database Server

Usage (HTML Interface) SGE Computation Grid WebService WebServer Server polling „BiBiServ“ „BiBiWSServ“ Database Server

Usage (HTML Interface) SGE Computation Grid WebService Result WebServer Server „BiBiServ“ „BiBiWSServ“ Database Server

Usage (HTML Interface) Result SGE Computation Grid WebService WebServer Server „BiBiServ“ „BiBiWSServ“ Database Server

Usage: other clients Perl Client SGE Computation Java Grid Client WebService Server „BiBiWSServ“ C# Client Database Server ??? Client

HTML Interface Usage 50.000 38.750 27.500 16.250 5.000 04 05 06 07 08 09 10 11 12 01 02 03 04 05 06 07 08 09 10 11 12 01 02 03 04 2005 2006 2007

WebService Interface Usage 4.500 3.375 2.250 1.125 0 04 05 06 07 08 09 10 11 12 01 02 03 04 05 06 07 08 09 10 11 12 01 02 03 04 2005 2006 2007

HOBIT • H elmholtz O pen B io I nformatics T echnology • cooperation of 11 german universities and Helmholtz Institutes as core of a network linking bioinformatics centres together • Goal: „To connect bioinformatics applications and resources in a uniform way to provide an efficient communication tier for resource access.“ • http://hobit.sourceforge.net/

Interoperation Obstacles • necessary prerequisite: common communication protocols & common data formats • most common bioinformatics data formats: • tool specific ASCII or binary files (but often used for other purposes) • no explicit structure • no or only implicit semantics (e.g. lower vs. upper case letters) • not easily extendable

Hobit Schemas • extendable XML Schema definitions for bioinformatics data • Basic building blocks for bioinformatical data types • Several higher-level types: • SequenceML • AlignmentML • RNAStructML • http://bioschemas.sourceforge.net/

BioDOM • Java Library for creation of data in HOBIT XML formats + conversion of existing legacy data types • easy adaptation of existing tools via: • BioDOM Library calls • usage of BioDOM WebService • http://biodom.sourceforge.net/

Workflow / Formats Example RNAStructAlignmentML RNA- Forester TinySeq mfold BioDOM BioDOM PNG/JPG/SVG RNA- >id BioDOM agtc Shapes uact RNAMovies SequenceML RNAStructML FASTA RNA- Fold 4SALE EMBLxml

BiBiServices as components • hand-coded workflows • WSDL-Description -> Taverna, Wildfire • jABC / Bio-jETI • Virtual BiBiServ

The (near) Future • Better description of services (WSDL is mostly OK, but not enough) • Semantics Layer • Integration with other Frameworks (e.g. BioMoby-S)

Why we are here... We are currently looking for generally accepted best practices and open standards for: • adding a semantic layer • better service descriptions • automated tool discovery methods • component interface requirements for different workflow design tools

Thanks for your attention :-)! Sven Hartmeier Faculty of Technology - Bielefeld University shartmei@techfak.uni-bielefeld.de http://bibiserv.techfak.uni-bielefeld.de/