What is Microarray Analysis?
08/12/2010 13:45
Microarray technology allows biologists to monitor genome-wide patterns of
gene expression in a high-throughput fashion. Applications of microarrays
have resulted in generating large volumes of gene expression data with several
levels of experimental data complexity. For example, a “simple” experiment
involving a 10,000-gene microarray with samples collected at five time points
for five treatments with three replicates can create a data set with 0.75 million
data points! Historically, hierarchical clustering was the first clustering
method applied to the problem of finding similar gene expression patterns
in microarray data. Since then many different clustering methods have been
used, such as k-means, a self-organizing map, a support vector machine,
association rules, and neural networks. Several commercial software packages,
e.g., GeneSpring or Spotfire, offer the use of these algorithms for microarray
analysis.
Today, microarray analysis is far beyond clustering. By incorporating
a priori biological knowledge, microarray analysis can become a powerful
method for modeling a biological system at the molecular level. For example,
combining sequence analysis methods, one can identify common promoter
motifs from the clusters of coexpressed genes in microarray data using various
clustering methods. Furthermore, any correlation among gene expression
profiles can be modeled by artificial neural networks and can hopefully
reverse-engineer the underlying genetic network in a cell.