Data mining for biomarker development: a review of tissue specificity analysis.

Novel biomarker development requires a significant resource commitment to translate candidate markers into clinical assays. Consequently, it is imperative high quality candidates are selected early in a biomarker development program. High throughput gene expression data are routinely used to identify transcripts differentially expressed in diseased versus normal samples. Data-mining Expressed Sequence Tag, Serial Analysis of Gene Expression, Massively Parallel Signature Sequencing, and microarray expression databases can provide additional information on the expression of candidate biomarkers across multiple tissues, organs, and disease states. From this information, quantitative measures of tissue-specific gene specificity are computed and used to guide candidate biomarker selection.