A fast algorithm for the exhaustive analysis of 12-nucleotide-long DNA sequences. Applications to human genomics

We have developed a new algorithm that allows the exhaustive determination of words of up to 12 nucleotides in DNA sequences. It is fast enough as to be used at a genomic scale running on a standard personal computer. As an example, we apply the algorithm to compare the number of all 12-nucleotide long words in human chromosomes 21 and 22, each of them more than 33 million nucleotides long. Sequences that are chromosome specific are detected in less than 2 minutes, being analyzed any pair of chromosomes at a rate of 45 millions of nucleotides (45 Mb) per minute. The size of the words is long enough as to allow further analyses of all significant sequences using conventional database searches. This allows to very simply establish the location and, many times, the biological meaning of the selected words. As an example, we show here, for the comparison between human chromosomes 21 and 22, that all the sequences that are found at least 40 times in one chromosome but are absent in the other belong to just two different classes, namely tandem repeats or genes with characteristic, internally repetitive, coding regions. Other available versions of this program and further applications are discussed.