Information Technology Reference
In-Depth Information
Tr i ana
Triana [318] is a workflow environment that has been developed at Cardiff
University, UK. It has in particular been designed to support building grid
workflows in physics and astronomy, but is in principle domain-independent
and has also been used to implement bioinformatics workflows (cf., e.g., [217]).
Triana provides a comprehensive set of ready-to-use, powerful data analysis
tools for tasks like signal analysis and image manipulation, and is extensible
to other, for instance, Web, grid, and P2P services, which reflects Triana's aim
of integration with grid technologies. The services are connected graphically
in order to define the data flow, but dedicated nodes for common control-flow
structures are provided in order to achieve basic support for additional control.
Wildfire
Wildfire [315] has been developed by A*STAR Bioinformatics Institute, Sin-
gapore. It serves as a graphical interface for defining GEL workflows. GEL
(Grid Execution Language) [186] is a scripting language for describing parallel
workflows, which can also be executed on Grids or cluster machines. Primar-
ily, Wildfire provides interfaces to the tools of the EMBOSS suite, but it is
possible to extend the component library to comprising further command-
line programs. Workflows are simply defined by connecting components with
pipelines that define the flow of data.
Wings
Pegasus is “a framework for mapping complex scientific workflows onto
distributed systems” [83] that has been used in a number of applications
from different scientific disciplines, also including bioinformatics. The Wings
(Workflow INstance Generation and Selection) system [111] that has been
fully integrated into Pegasus provides functionality for (semi-) automatic
workflow creation based on semantic representations and planning techniques.
Wings puts a strong focus on the analyzed data, that is, intermediate results
can influence the generation of subsequent workflow parts, and semantic de-
scriptions of results are automatically generated. The system distinguishes
three distinct stages in workflow development, namely the creation of work-
flow templates (where an expert user specifies high-level, data-independent
workflow templates), the creation of workflow instances (where the normal
user specifies the data that is to be used by the computation) and the cre-
ation of the actually executable workflows (where the workflow instance is
automatically mapped to a concrete execution environment by Pegasus).
Indeed, Wings supports loose specification, concretization, and also
constraint-based validation of workflows in a way that reminds of the loose
programming approach followed by Bio-jETI/PROPHETS. However, within
Wingsthelooselyspecifiedworkflowtemplatesaswellasallworkflowcon-
straints are already defined as part of the domain model by an expert user,
 
Search WWH ::




Custom Search