Running time of LTR_FINDER is very sensitive to this parameter. The
program only selects pairs that are longer than this value to do
further processes from all exact match pairs detected. If a very
small value is given, LTR_FINDER will spend much time on randomly
matched short sequences. We use P-value to estimate proper value of
-p, that is, the probability of exact match of length longer than
occurs if 2 sequences are drown randomly. In Waterman
1989(, ), under independent letter model, using
asymptotic extreme value distribution, P-value were worked out
analytically. Now if one assign the P-value, length can be deduced.
From experiments, 20 is appropriate in most situation, and we
suggest not using of value less than 15.

2009-04-09