Essentially all diseases and metabolic states are polygenic. Even if a physical state were controlled by a single gene coding for a single protein, the up regulation of that gene would perturb the broader system, and many coding regions would be affected. As a result, the genome-centric view of molecular biology is slowly being replaced by a more comprehensive systems view. One of the most important elements of this approach is a complete understanding of the transcriptional state of all genes involved in a specific metabolic state. As a result, technologies for gene-expression profiling have become central to molecular biology and diagnostic medicine.
Any technique used to study the transcripts within a cell must be capable of spanning the range from single-digit copy counts to very large numbers, often in the thousands. Accuracy is important because at the single-digit level, small changes in the number of copies of certain messages can have significant effects on metabolism and disease. This discussion focused on various technologies for transcriptional profiling, including both physical techniques and mathematical algorithms. Because different technologies are appropriate for different types of experiments, this chapter presented a balanced view that does not particularly favor any one approach. Likewise, this review of the basic classes of clustering algorithms is designed to highlight different approaches with a focus on the theoretical differences. Moreover, readers are encouraged to consider the differences presented here when reviewing experimental results. These differences are intended to form a backdrop for the next phase of our discussion, which focuses on protein translation.