An equivalence test for comparing DNA sequences

Abstract

The practice of sequence alignment is constantly oscillating between the risk of overlooking important structure and that of discovering any arbitrarily defined kind of structure anywhere. On the other hand, the use of a condensed consensus sequence may lead to a substantial loss in valuable information. While adopting a Mahalanobis-type index we allow for a certain degree of uncertainty in the measurements. This uncertainty may be caused by inaccurate measurements or ambiguity. In this paper, we test the similarity between DNA sequences within the framework of equivalence testing, accounting for both variances and covariances between frequencies of nucleotides. Statistical methods for testing equivalence were first developed in the context of pharmacokinetics and later extended to the field of clinical trials. Nowadays, (bio)equivalence tests seem to be less frequently used outside the drug testing field, including statistical genetics. Copyright (c) 2005 John Wiley & Sons, Ltd.