Information Technology Reference
In-Depth Information
7.2.3 Exemplary caGrid Workflows
This section shows two caGrid Workflows we have created and
uploaded to myExperiment. The first workflow is a simple task, that
is, querying and fetching microarray data; while the second one is a
fully fledged routine with real scientific value, that is, lymphoma type
prediction. For all the caGrid workflows and other Taverna ones, users
are encouraged to visit the link
http://www.myexperiment.org/search?
query
workflows
. All caGrid workflows are modeled in a
graphical way and organized into meaningful modules, such that they
can easily be reused out of the box or be modified and repurposed.
¼
cabig&type
¼
Workflow 1: Access to caArray Data Using
cql-builder and caGrid-transfer
caArray [195] is an open-source, Web and programmatically accessi-
ble microarray data management system developed by caBIG. Since it
acquires, disseminates, and aggregates a large volume of cancer-
related array data, cancer researchers in caBIG frequently start their
in-silico
investigation by querying and retrieving data from caArray
and subsequently analyze them using tools and services on and off the
caGrid.
Figure 7.8 shows a workflow that queries all the files related to a
microarray experiment and selects and downloads some of them. The
input of the workflow is the identifier of the microarray experiment
(
experiment_id
) of interest. Within the CQL builder (
CQL_Builder
), a
complex CQL clause is built, querying all the caArray file objects
associated with this experiment as shown via the CQL builder GUI and
the criteria editing dialog in Figure 7.9. A user can then pick up one or
several caArray files (
extract_a_file
), create a download session
(
createFileTransfer
), and use the caGrid-transfer-activity plug-in
(
caGrid_Transfer_Activity
) to download files to a local directory.
The output is the name of the downloaded file (
resultFile
). From
now on, we only mention the key activities/services in a workflow,
and those that are not mentioned may be local activities doing data
transformations, XML manipulations, and so on.
In the CQL builder GUI (the upper dialog of Figure 7.9), users can
choose which service to query by giving the service's URL, and then, its
object-oriented data model is populated and ready to be selected. Users
can select one of them (
gov.nih.nci.caarray.domain.file.CaArrayFile
)
