The cardinal argument for laser-microdissection-based expression profiling is, however, not technical, but analytical. Laser microdissection allows the user to generate expression data in a defined cellular context. More often than not, the significance of expression of a particular gene depends on the identity of the cell harboring it. Perhaps the biggest challenge in microarray experiments lies in the analysis of the data generated. Gene-by-gene analysis is time consuming, relies on presumed functions of genes, and cannot capture complex phenomena. Examples of going beyond gene-by-gene analysis are mapping of expression data onto pathways, such as those provided by Ingenuity™ (Mountain View, CA) and Jubilant Biosys (Columbia, MD), pattern matching (Hughes et al., 2000), and network analyses (Tavazoie et al., 1999; Wille et al., 2004). A discussion of microarray data analysis is beyond the scope of this chapter, and the reader is referred to several recent reviews (Sherlock, 2001; de la Fuente et al., 2002; Kaminski and Friedman, 2002; Hariharan, 2003; Leung and Cavalieri, 2003).
Some of the more advanced types of analyses use expression data as a snapshot of the cell "system." To ensure that this snapshot represents a particular state of the cell system, the cell sample must be homogeneous. Accordingly, much of the pattern-matching and network-modeling analyses has been performed on cell culture, or yeast, data. The application of these analysis tools to expression data from in vivo models is needed, and laser microdissection will be required to provide the necessary cell-sample homogeneity from many tissues. Sample homogeneity also clearly increases the possibility of seeing coordinate regulation of genes within a given pathway. This opens new statistical possibilities. When several genes within an a priori defined pathway are tested, the statistical probability that the pathway is regulated is a function of all the parts of the pathway. Thus, several genes that are modestly, and statistically insignificantly, regulated may together show that a shared pathway is significantly regulated (Mootha et al., 2003). Due to the large number of data points, microarrays may actually be more sensitive than quantitative PCR of a smaller number of genes for detecting pathway regulation in such a fashion.
Was this article helpful?