Avoiding Ambiguity and Assessing Uniqueness in Minisatellite Alignment

Several algorithms have been suggested for minisatellite alignment. Their time complexity is high—close toO(n3)—due to the necessary reconstruction of duplication histories. We investigate the uniqueness of optimal alignments computed under the common single-copy duplication model. To this extent, it is necessary to avoid ambiguity in the algorithm employed. We re-code the ARLEM algorithm in the form of a grammar, and apply a disambiguation technique which uses a mapping to a canonical representation of minisatellite alignments. Having arrived at a nonambiguous algorithm this way, we demonstrate that the underlying model—independent of the algorithm—gives rise to an exorbitant number of different, co-optimal alignments when applied to real-world data. We conclude that alignment-free methods should be considered for minisatellite comparison. 1998 ACM Subject Classification J.3.a [Life and Medical Sciences]: biology and genetics