A novel changepoint statistic based on the minimum value, over possible cha
ngepoint locations, of Fisher's Exact Test, is introduced. Specific points
in the exact distribution of the minimally selected Fisher's value may be r
apidly calculated as a lattice-path counting problem via known recurrence m
ethods. The test is compared to the Kolmogorov-Smirnov two-sample test, the
maximally selected chi-square, and a likelihood ratio test. The tests are
applied to assessing recombination in genetic sequences of HIV.