Biology Reference
In-Depth Information
To solve the problem of mining microarray expression data, GAs are
especially useful for the following reasons:
•
The problem space is large and complex.
•
Prior knowledge is scarce.
•
It is difficult to determine a machine learning model to solve the prob-
lem due to complexities in constraints and objectives.
•
Traditional search methods, such as stochastic, combinatorial, and clas-
sical (so-called “hard”) optimization-based techniques, perform badly.
4.4.2.
SDL global optimization algorithms
Although GAs are popular and useful, many problems at hand cannot be
resolved easily and accurately. This section combines a powerful algo-
rithm (Li, 2004), which has been used in optical coating design, with the
methods of cancer diagnosis through gene selection and microarray analy-
sis. We name this novel analytical method of DNA microarray data “SDL
global optimization”. A generic approach to cancer classification based on
gene expression monitoring by DNA microarrays is proposed and applied
to a test leukemia case. By using the orthogonal arrays for sampling and
a search space reduction process, a computer program has been written
that can operate on a personal laptop computer. The leukemia microarray
data can be classified 100% correctly without previous knowledge of their
classes.
Search WWH ::
Custom Search