On the detection of DNA periodicities by spectral analysis and adaptive mappings: Cost functions comparison

DNA sequences can be analyzed by spectral techniques, looking for hidden periodicities. Anyway, mapping between nucleotide bases and numeric values has a strong impact on the DNA spectrum. The Minimum Entropy Mapping algorithm introduced the concept of adaptive mapping: the association between bases and numbers is not fixed, but is chosen to minimize a cost function, the spectrum entropy. In this paper, we consider adaptive mappings based on other cost functions, like Simpson Variance, Gini Difference and Bray Measure. Their different ability to detect periodicities and exonic regions is addressed and discussed.