Biomedical Engineering Reference
In-Depth Information
information resides in databases following this model. For example, a descendent of MUMPS called
simply M is the standard for EMRs in the Veterans Administration hospitals throughout the U.S.
Because of the storage inefficiency of the hierarchical model for some types of data, the network
model was developed in the late 1960s. For example, the network model is more flexible than the
hierarchical one because multiple connections can be established between files. These multiple
connections enable the user to gain access to a particular file more effectively, without traversing the
entire hierarchy above that file. Unlike the one-to-many relationship supported by the hierarchical
model, the network model is based on a many-to-one relationship. The network model is significant
in bioinformatics in that it may play a significant role in the architecture of the Great Global Grid and
other Web-based computing initiatives.
One of the most significant alternatives to the relational database model is the object-oriented model
in which complex data structures are represented by composite objects, which are objects that
contain other objects. These objects may contain other objects in turn, allowing structures to be
nested to any degree. This metaphor is especially appealing to those who work with bioinformatics
data because this nesting of complexity complements the natural structure of genomic data (see
Figure 2-18
).
Figure 2-18. Object-Oriented Data Representation. The object-oriented data
model is natural for hiding the complexity of genomic data.
The object-oriented model combines the natural structure of the hierarchical model with the flexibility
of the relational model. As such, the major advantage of the object-oriented model is that it can be
used to represent complex genomic information, including non-record-oriented data, such as textual
sequence data and images, in a way that doesn't compromise flexibility. Furthermore, with an object-
oriented DBMS, it's possible to use arbitrary data types, and complex relationships can be queried
without having to create resource-intensive joins between tables. The object-oriented model is
considered optimum for handling genomic data, because it allows combinations of data to be treated
as single entities. Instead of thinking about a gene with exons, introns, mRNA, nucleotide sequences,
associated proteins, and their 3D shapes as a separate sound file, a separate video file, and a
