Effective analysis of genomic data

High-throughput biotechnology has enabled genome-wide investigation of gene expression and has the potential to identify genes that have a role to play in focal cerebral ischemia, as well as many other interventions. The advent of this technology has also led to the generation of large amounts of expensive and complex expression data. One of the major problems with the generation of so much data is locating and extracting the relevant information to aid target identification and interpretation effectively and reliably. Statistical involvement is vital. Not only does it help to ensure effective extraction of information from the data, it also increases the likelihood that the data collected will embody the information about the differential expression of interest in the first place. The goal of this chapter is to recommend an effective process for investigating gene expression data. There are five stages in this process that we believe lead to reliable results when routinely applied to an expression dataset, once it has been appropriately generated and collected: (1) biological problem definition and design selection; (2) data examination, "preprocessing," and reexamination; (3) data analysis step I: screening for differentially expressed genes; (4) data analysis step II: verifying differential expression; and (5) biological verification, interpretation, and communication.