“…This model hasl = 25 and δ l = 11.2. There is nothing special about the Erlang function except that it was a simple generalize function of the exponential functionthat we used in the first instnace -and that it provided a simple setting for discrete parameter search: after we realizedl needed to be less than 50, we scannedl in intervals of 5 and n in intervals The following are some examples that gave very good k-distributions for specific k-mers but not generally; all were generated with L 0 = 1000 and n = 0: for 6-mer, σ = 13, 000 ± 2, 000 and on average 0.04σ mutations per duplication (these parameters also work for genomes with biased base compositions) [14]; for 2-mer, σ = 50, no mutation; for 5-mer, σ = 30, no mutation; for 9-mer, σ = 15, no mutation.…”