Abstract
This chapter describes the system and method termed as dragon promoter program. It aims at producing both high accuracy predictions of vertebrate promoters and high positional accuracy of TSS predictions. The DRAGON allows users to select among five specificity levels of predictions, ranging from a very high specificity to a very high sensitivity. Very high sensitivity predictions are suitable for screening relatively short fragments of a genomic sequence, whereas at the other extreme, very high specificity predictions are suitable for genome-scale predictions because of the significantly reduced number of false-positive predictions. It is expected that improvements in computational promoter analysis will follow two paths. First, the increased number of experimentally verified promoters and their TSSs will help further refine promoter prediction methods and increase the overall accuracy. Second, the improved analysis of promoter structure will help understand the diversity of promoters and identification of specific patterns that regulate gene expression. The analysis of promoters will remain a critical issue in the analysis of genes and genomes, particularly for the study of regulation of gene expression and gene networks. Because of the large size of vertebrate genomes, computational analysis will remain the key technology in the analysis of functional sites in DNA.