PROPERTIES OF THE SIMILARITY INDEX FROM DNA FINGERPRINTING DATA

                            Stephen Haslett
                  Statistics Research and Consulting Centre
                       Massey University, New Zealand


For DNA profiles taken from different individuals the similarity index has
been suggested as a proxy for genotypic identity-in-state and population
homozygosity.
Similarity within a population can be measured by the average proportion of
matching bands from the DNA profiles for all possible pairwise comparisons
between individuals. Lynch (1990) considered some initial properties of
this index, but because the data for the index are comparisons between
pairs of individuals, the correlation between data points introduces a
complication in the estimation of standard errors for the similarity index.
 A parallel problem arises when comparing populations.  This seminar
considers estimation of standard errors, and the links of this problem to
analysis of variance and graph theory. The problem is generalisable to any
situation where raw data, however measured, involve all possible pairwise
comparisons. Parts of the analysis will be illustrated with some data
collected from humpback whales from four different locations: Gulf of
Maine, Southeast Alaska]a, California, and Hawaii.

Lynch, M (1990) The similarity index and DNA fingerprinting, Molecular
Biology and Evolution, 7(5) 478-484.