Biomedical Engineering Reference
In-Depth Information
Oracle server (Oracle version 11.2) supporting research class applications.
Typically, two environments are supported as described in the previous
section. The netCDF data fi les and indexes are managed on general NAS
fi lers that are available to the application hosts, this is where the bulk of
the data resides (the database size is small being a few GB). The Tomcat
web application is deployed on an Enterprise Red Hat Linux system, (HP
Proliant DL 580 with 4 CPUs and 64 GB memory) running under a
locally installed Java Runtime Environment.
A corporate system that includes proprietary data should have access
control. In the Pfi zer deployment, login and authentication was
implemented by developing a standard J2EE-based request fi lter that
integrates with a single sign-on system used throughout the company.
The fi lter class and supporting libraries are simply installed to Tomcat's
standard directories for shared libraries or classes. The fi lter is referenced
in the web application deployment fi le, so no custom extensions or
modifi cations are required.
A couple of minor extensions have been developed to date to facilitate
the Pfi zer deployment. First, it was desired to add a corporate logo to
the in-house installation to distinguish the site from the public Atlas.
This was done initially through modifi cation of an application JSP page.
The EBI Atlas team then developed a template model that allows the
site to be 'branded' for local installations without modifi cation of
source code. A minor code change has also been implemented to
facilitate capturing of application usage metrics, so code for the system is
currently obtained from the source-code control system supporting
the open source project and used to compile a custom build containing
the local customizations.
9.4.3 Loading custom and private data sets
into the in-house Atlas
Although the EBI public Atlas contains thousands of curated public
experiments, the wealth of public data submitted historically and
currently will require local curation to include additional studies of
interest. In addition, public and private research institutions will have
access to internally generated and proprietary data to add to their local
Atlas installation. Therefore, an institution interested in setting up an
in-house Atlas instance will want to consider workfl ow and resource for
curation of internal data. The Pfi zer installation of Gene Expression Atlas
has been supplemented with public studies of interest to scientists and
Search WWH ::
Custom Search