Biomedical Engineering Reference
In-Depth Information
The flat data model is simply a table without any embedded structure information to govern the
relationships between records. As a result, a flat file database can only work with one table or file at
a time. Strictly speaking, a flat file doesn't really fit the criteria for a data model because it lacks an
embedded structure. However, the lack of an embedded structure is one reason for the popularity of
the flat file database in bioinformatics, especially in capturing sequence data. A sequence of a few
dozen characters may be followed by a sequence of thousands of characters, with no known
relationship between the sequences, other than perhaps the tissue sample or sequence run. As such,
a separate flat file can be used to efficiently store the sequence data from each sample or run. In
order to make the management of large amounts of sequence or other data more tenable, a model
with an embedded structure is required.
The relational model, developed in the early 1970s, is based on the concept of a data table in which
every row is unique. The records or rows in the table are called tuples; the fields or columns are
variably referred to attributes, predicates, or classes. Database queries are performed with the select
operation, which asks for all tuples in a certain relation that meet a certain criterion—for example, a
query such as "Which authors write about neurofibromatosis?" To connect the data of two or more
relations, an operation called a join is performed. A record is retrieved from the database by means
of a key, or label, that may consist of a field, part of a field, or a combination of several fields.
Supporting this data model so that it's easy for someone to direct a search for the record that
contains the particular value of the key is the purpose of a relational DBMS. Consider querying a
bibliographic database with an "author_subject_table," using the Structured Query Language (SQL)
statement:
SELECT *.* FROM author_subject_table
WHERE subject = "Neurofibromatosis"
A useful feature of the relational model is that records or rows from different files can be combined as
long as the different files have one field in common. Theoretically, records with a common field can
be combined or joined with an unlimited number of files. The price paid for this flexibility is extended
