The Origin and History of in silico Experiments

1
The Origin and History of in silico
Experiments Jun Zhao, Robert D. Stevens, Chris
J. Wroe, Mark Greenwood and Carole A.
Goble Dept. of Computer Science, University of
Manchester, M13 9PL
Process level, collects how, when and where the
workflow is run, what data are used and
generated, which computational services are
invoked, and the input and output data for
service invocation
Introduction
It is not enough to be able to just run an
e-Science in silico experiment it is also vital
to be able to understand and interpret the
outputs of those experiments. The results have
little value if other scientists, or even the
same scientist at a later date, are unable to
identify their origin, or provenance. myGrid is
a UK e-Science pilot project which is developing
Grid middleware infrastructure for in silico
experiments in biology. myGrid regards each in
silico experiment run as a workflow. In the
myGrid project, we model provenance to provide
the most common and expected context (what,
which, why, when, where and who), resource
(where), and derivation (how) information about
in silico experiments.
Data level, primarily inferred from the process
level provenance, describes the derivation path
of the data final and intermediate results
Organisation level, records the workflow user and
creator, their organisation, project, the
hypothesis for this experiment/project, the
experiment design, etc Knowledge level, links
the knowledge outcome of the workflow with the
process, data and organisation provenance that
provide the evidence to support it.
Four-level Provenance Pyramid
Provenance Generation and Browsing
Provenance generation is supported by the
provenance data model, part of the myGrid
information model, which provides a standard by
which to structure information about
bioinformatics experiments. This process, data
and organisation provenance is automatically
generated when creating and running workflows
with the Taverna toolkit, including the Freefluo
workflow enactment engine, as shown in the
following figure on the left. The Life Sciences
Identifier (LSID) scheme is used to uniquely and
persistently identify and resolve a data resource
and its associated metadata. The Resource
Description Framework (RDF) is used to represent
the provenance graph, with LSIDs as the Unique
Resource Identifiers (URIs) to link resources
together. In order to present a rich view of
this RDF provenance graph, Haystack is applied.
The following figure on the right shows an
example RDF graph viewed in Haystack, which
displays how resources are linked together in
four different views (process, data, organisation
and knowledge views).
Example Workflow Run from myGrid.
The RDF Provenance Graph Viewed in Haystack.
Acknowledgements The authors would like to
acknowledge the myGrid team Matthew Addis, Nedim
Alpdemir, Andy Brass, Rich Cawley, Tracy
Craddock, Neil Davis, Dave De Roure, Alvaro
Fernandes, Justin Ferris, Robert Gaizauskas,
Kevin Glover, Carole Goble, Chris Greenhalgh,
Mark Greenwood, Yikun Guo, Simon Harper, Keith
Hayward, Ananth Krishna, Peter Li, Phil Lord,
Darren Marvin, Karon Mee, Simon Miles,
Luc Moreau, Arijit Mukherjee, Tom Oinn,
Steve Oliver, Juri Papay, Savas Parastiditis,
Norman Paton, Milena Radenkovic, Peter Rice, Tom
Rodden, Martin Senger, Nick Sharman, Robert
Stevens, Victor Tan, Paul Watson, Anil Wipat and
Chris Wroe. The Williams-Beuren Syndrome case
study is from Hannah Tipney, May Tassabehji, Andy
Brass, St Marys Hospital, University of
Manchester, UK
Provenance Generation in myGrid.

• References
• http//www-124.ibm.com/developerworks/oss/lsid/
• http//haystack.lcs.mit.edu/
• Taverna workbench. http//taverna.sourceforge.net
  
• Freefluo engine. http//freefluo.sourceforge.net