Biomedical Engineering Reference
In-Depth Information
Figure 9.4
Atlas architecture
simplifying the underlying data model and creating a new framework for
loading data. The overall architecture is described in Figure 9.4 below.
9.3.2 Simplifying the data model
To simplify maintenance and data administration, we moved from a highly
formalized data structure, where every attribute was encoded explicitly as
a separate set of tables, into a more lightweight data model, fewer than 30
tables in total. This also allowed us to simplify greatly the data loading
procedure, going almost directly from MAGE-TAB parser objects to Atlas
data objects and then to the relational database. The database itself does
not store the raw data or analytics: these are stored in NetCDF format [13]
fi les and accessed directly by the web application and R-based analytics.
These NetCDF fi les are also available for power users to access via R
scripts, if they need to do more sophisticated processing of the raw data.
9.3.3 Building an easy to use installation
procedure
The original Atlas was not meant to be installed anywhere except at the
EBI. This meant there were a number of dependencies on existing
infrastructure. We developed a regular software and data release
Search WWH ::
Custom Search