Because paired-end genomic signature tags are sequenced-based, they have the potential to become an alternate tool to tiled microarray hybridization as a method for genome-wide localization of transcription factors and other sequence-specific DNA binding proteins. As outlined here the method also can be used for global analysis of DNA methylation. One advantage of this approach is the ability to easily switch between different genome types without having to fabricate a new microarray for each and every DNA type. However, the method does have some disadvantages. Among the most rate-limiting steps of our PE-GST protocol are the need to concatemerize the diTAGs, size fractionate them and then clone them prior to sequencing. This is usually followed by additional steps to amplify and size select for long (> or = 500) concatemer inserts prior to sequencing. These time-consuming steps are important for standard DNA sequencing as they increase efficiency approximately 20-30-fold since each amplified concatemer can now provide information on multiple tags; the limitation on data acqui- sition is read length during sequencing. However, the development of new sequencing methods such as Life Sciences' 454 new nanotechnology-based sequencing instrument (41) could increase tag sequencing efficiency by several orders of magnitude (> or = 100,000 diTAG reads/run), which is sufficient to provide in-depth global analysis of all ChIP PE-GSTs in a single run. This is because the lengths of our paired-end diTAGs (approximately 60 bp) fall well within the region of high accuracy for read lengths on this instrument. In principle, sequence analysis of diTAGs could begin as soon as they are generated, thereby completely bypassing the need for the concatemerization, sizing, downstream cloning steps and sequencing template purification. In addition, our protocol places any one of several unique four-base long nucleotide sequences, such as GATC, between each and every diTAG pair, which could be used to help the instrument's software keep base register and also provide a well-located peak height indicator in the middle of every sequence run. This additional feature could permit multiplexing of the data by simultaneous sequencing of several pooled libraries if each used a different linker sequence during diTAG formation (Figure 4).