Abstract
Molecular indexing sorts DNA fragments into subsets for inter-sample comparisons. Type IIS or interrupted palindrome restriction endonucleases, which result in single-stranded ends not including the original recognition sequence of the enzyme, are used to produce the fragments. The ends can then be any sequence but will always be specific for a given fragment. Fragments with particular ends are selected by ligation to a corresponding indexing adapter. We describe iterative indexing, a new process that after an initial round of indexing uses a Type IIS restriction endonuclease to expose additional sequence for further indexing. New plasmids, pINDnn, were produced for novel use as indexing adapters. Together, the plasmids index all 16 possible dinucleotides. Their large size can be increased by dimerisation in vitro and allows the isolation of indexed material by size separation. Fragments produced from human genomic DNA by Type II restriction endonucleases were sorted using six bases in total to a possible enrichment of 1920-fold. By comparison with the public human sequence databases, fidelity of indexing was shown to be high and was tolerant of repetitive sequences. Genome-wide comparisons on a candidate or non-candidate basis are made possible by this approach.